chore(release): v3.24.0 — OpenAI multimodal + structured outputs, SPOT prompt budget, agentic fix, OCP_LOCAL_TOOLS (#186)

* chore(release): v3.24.0 — OpenAI multimodal vision + structured outputs, SPOT prompt budget, agentic-turn fix, OCP_LOCAL_TOOLS

Consolidates #179/#153/#154/#183/#181/#182 (merged since the v3.23.0 tag).
3.23.0 → 3.24.0.

Minor: four user-facing features (multimodal vision #154, structured outputs
#153, SPOT-derived prompt budget #179, OCP_LOCAL_TOOLS #182) + two fixes
(#183 agentic final-answer, #181 deep-reply-500). No breaking change; no new
endpoint; no new cli.js wire behavior. Two features from @vvlasy-openclaw.

Release-kit walk: new env vars all documented in-PR (CLAUDE_IMAGE_ALLOW_URL /
CLAUDE_MAX_IMAGE_* #154, OCP_STRUCTURED_MAX_ATTEMPTS #153, OCP_LOCAL_TOOLS #182 —
grepped present in README env table); ADR 0009 (#179) + ADR 0006 (B.1, #153/#154)
indexed; version from package.json only (no stale refs); the CHANGELOG Unreleased
section (which held #182) is retitled to v3.24.0 with the other five entries added.

Tag push v3.24.0 at this squash commit triggers release.yml.

Co-Authored-By: Claude <claude-opus-4-8> <noreply@anthropic.com>

* docs(changelog): correct contributor count — four of six PRs are @vvlasy-openclaw's, not two (release reviewer)

---------

Co-authored-by: dtzp555 <dtzp555@gmail.com>
Co-authored-by: Claude <claude-opus-4-8> <noreply@anthropic.com>

.github/PULL_REQUEST_TEMPLATE.md

@@ -4,22 +4,46 @@
 
 <!-- One or two sentences describing the change and why it is in scope for OCP. -->
 
+## Endpoint Class (REQUIRED)
+
+Per `ALIGNMENT.md` and ADR 0006, every PR that touches a network-facing endpoint must declare its class. Pick the most specific applicable class (Hybrid covers PRs that touch both A and B):
+
+- [ ] **Class A** — forwards a `cli.js` operation (e.g., `/v1/messages`, `/api/oauth/*`, or the Anthropic-side wire call inside `/usage`)
+- [ ] **Class B** — extends an OCP-owned compatibility endpoint (per ADR 0006). Sub-bucket:
+  - [ ] B.1 — OpenAI-compatibility surface (`/v1/chat/completions`, `/v1/models`)
+  - [ ] B.2 — OCP-administrative surface (`/health`, `/dashboard`, `/sessions`, `/logs`, `/status`, `/settings`, `/api/keys*`, `/api/usage`, `/cache*`)
+- [ ] **Hybrid** — touches both classes (e.g., `/usage` if the PR modifies both the Anthropic wire call AND the local synthesis layer). Both evidence sections below must be filled.
+- [ ] **Not endpoint-touching** — refactor / docs / tooling that does not modify any request handler. Skip both evidence sections; explain in Summary.
+
 ## Claude Code Alignment Evidence (REQUIRED)
 
-Per `ALIGNMENT.md`, every PR that touches `server.mjs` or any network-facing surface must fill out this section. PRs with this section blank or unchecked will receive a `request changes` review and cannot be merged.
+PRs with the relevant evidence section blank or unchecked will receive a `request changes` review and cannot be merged.
+
+### If Class A
 
 - [ ] **Corresponding `cli.js` reference.** I have identified the `cli.js` function and line range that performs the operation this PR forwards. Citation (format `cli.js:NNNN` or `cli.js vE4 <functionName>`):
   <!-- e.g. cli.js:18423-18467 (function: sendUserMessage)  -->
 
-- [ ] **If `cli.js` does not perform this operation**, I have stated this explicitly below and justified the scope under `ALIGNMENT.md` Rule 2. (Note: in almost all cases this means the PR should be closed, not merged. Proxy layers do not invent endpoints.)
+- [ ] **If `cli.js` does not perform this operation**, I have stated this explicitly below and justified the scope under `ALIGNMENT.md` Rule 2. (Note: in almost all cases this means the PR should be closed, not merged. Proxy layers do not invent endpoints. If the endpoint is in fact Class B, switch the class above and use the Class B section instead.)
   <!-- Justification, if applicable. Empty is fine when cli.js does perform the operation. -->
 
 - [ ] **Commit message citations.** Every "Claude Code uses X" or "cli.js uses X" assertion in every commit of this PR is immediately followed by a `cli.js:NNNN` or `cli.js vE4 <functionName>` citation. I have verified this by rereading each commit message.
 
+### If Class B
+
+- [ ] **Authorizing ADR.** Cite the ADR number that authorizes the endpoint this PR modifies (e.g., "ADR 0006 — OpenAI shim scope"). For B.1 endpoints (`/v1/chat/completions`, `/v1/models`), this is ADR 0006. For grandfathered B.2 endpoints, this is "ADR 0006 (grandfathered as of v3.16.4)." For new B.2 endpoints, cite the endpoint's own authorizing ADR; if none exists, the PR cannot proceed — the authorizing ADR must be drafted and merged first.
+  <!-- e.g., ADR 0006 -->
+
+- [ ] **Specification citation.** For B.1 endpoints, link to the relevant section of OpenAI's `/v1/chat/completions` specification (https://platform.openai.com/docs/api-reference/chat/create), including the specific field or behaviour being implemented. For B.2 endpoints with their own ADR, cite the ADR section that specifies the behaviour. For grandfathered B.2 endpoints, the PR must be a behaviour-preserving refactor — link the existing handler code being modified.
+  <!-- B.1 example: OpenAI chat/completions, `response_format` parameter, https://platform.openai.com/docs/api-reference/chat/create#chat-create-response_format -->
+  <!-- B.2 example: ADR 00NN § "Behaviour" -->
+
+- [ ] **No invention beyond the specification.** I confirm this PR does not introduce any field or behaviour not present in OpenAI's spec for the endpoint (B.1) or beyond the scope of the authorizing ADR (B.2). For grandfathered B.2 endpoints, I confirm the change is behaviour-preserving (no contract drift). If something the user actually wants is not in the spec, the right answer is to close this PR and propose an upstream spec change or a new ADR.
+
 ## Type of change
 
-- [ ] Bug fix (alignment with existing `cli.js` behavior)
-- [ ] Feature (new `cli.js` behavior now surfaced through OCP)
+- [ ] Bug fix (alignment with existing `cli.js` behavior, or with the cited spec / ADR for Class B)
+- [ ] Feature (new `cli.js` behavior now surfaced through OCP, or new field already in OpenAI's spec for Class B)
 - [ ] Refactor (no wire-level behavior change)
 - [ ] Deletion (unalignable feature removal per `ALIGNMENT.md` Unalignable Policy)
 - [ ] Documentation / governance
@@ -28,14 +52,16 @@ Per `ALIGNMENT.md`, every PR that touches `server.mjs` or any network-facing sur
 
 Reviewers: this section is for you, not the author. Do not approve until every box is checked.
 
-- [ ] I opened `cli.js` at the cited line range and confirmed the operation matches.
+- [ ] If Class A, I opened `cli.js` at the cited line range and confirmed the operation matches. If Class B, I opened the OpenAI spec at the cited section (B.1) or the authorizing ADR (B.2) and confirmed the behaviour described in this PR matches the cited reference.
 - [ ] I ran (or confirmed CI ran) `.github/workflows/alignment.yml` and it passed.
 - [ ] I am not the commit author of any commit in this PR (Iron Rule 10).
-- [ ] If the PR asserts scope without a `cli.js` citation, I confirmed the justification is sound per `ALIGNMENT.md` Rule 2.
+- [ ] If the PR asserts scope without a `cli.js` citation (Class A) or without an ADR (Class B), I confirmed the justification is sound per `ALIGNMENT.md` Rule 2 and ADR 0006.
+- [ ] If the PR is Class B and adds a new endpoint or new method, I confirmed the authorizing ADR lands in the same merge or before this PR.
 
 ## Related
 
-- `ALIGNMENT.md` Rule(s) invoked: <!-- e.g. Rule 3 -->
+- `ALIGNMENT.md` Rule(s) invoked: <!-- e.g. Rule 3, or Rule 3 (Class B mapping) -->
+- Authorizing ADR (Class B only): <!-- e.g. ADR 0006 -->
 - Related issue / prior PR: <!-- #NNN -->
 - Historical lesson reference (if relevant): <!-- e.g. 2026-04-11 drift, b87992f -->
 

.github/workflows/alignment.yml

@@ -4,6 +4,11 @@ on:
   pull_request:
     paths:
       - 'server.mjs'
+      - 'setup.mjs'
+      - 'scripts/**'
+      - 'lib/**'
+      - 'ocp'
+      - 'ocp-connect'
       - '.github/workflows/alignment.yml'
 
 jobs:
@@ -24,10 +29,14 @@ jobs:
             exit 0
           fi
 
-          # Known-hallucinated tokens. Extend only via an ALIGNMENT.md amendment PR.
-          # Each token is matched as a fixed string against server.mjs only.
+          # Blacklisted tokens — two kinds (see ALIGNMENT.md "OAuth token-host verification"):
+          #   (1) known LLM hallucinations (e.g. the 2026-04-11 /api/oauth/usage drift), and
+          #   (2) pinned wrong-host variants of a VERIFIED Class A endpoint (a hit means a
+          #       drift to a known-wrong host, not necessarily a hallucination).
+          # Extend only via an ALIGNMENT.md amendment PR. Matched as fixed strings vs server.mjs.
           BLACKLIST=(
             "api.anthropic.com/api/oauth/usage"
+            "console.anthropic.com/v1/oauth/token"
           )
 
           FAIL=0
@@ -46,8 +55,8 @@ jobs:
           ============================================================
           server.mjs contains a token on the OCP alignment blacklist.
 
-          These tokens were introduced by LLM hallucinations and do
-          not appear in cli.js at any shipped Claude Code version.
+          These tokens are either LLM hallucinations that never appeared in cli.js,
+          or pinned wrong-host variants of a verified Class A endpoint (a drift).
           See ALIGNMENT.md -> "Historical Lesson: The 2026-04-11 Drift"
           (commit b87992f) for the full incident record.
 
@@ -66,6 +75,80 @@ jobs:
 
           echo "Blacklist scan clean."
 
+  port-spot:
+    name: port literal SPOT (hard fail)
+    # Background: from 2026-05-08 (PR #71 dogfood accident) through 2026-05-13
+    # a hardcoded "3478" in scripts/upgrade.mjs + scripts/doctor.mjs cascaded
+    # into wrong baseUrl writes for the OpenClaw "claude-local" provider,
+    # taking out the "大内总管" Telegram agent.
+    #
+    # Rule: the only places allowed to write a literal port number in source
+    # are (a) lib/constants.mjs (the SPOT), (b) bash scripts ocp / ocp-connect
+    # (which can't import .mjs and must keep the literal in sync — flagged
+    # with a `// keep in sync with lib/constants.mjs` style comment), and
+    # (c) test-features.mjs (intentionally pins historical ports for plist /
+    # systemd parser tests). Everything else MUST import from lib/constants.mjs.
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Scan for hardcoded port literals outside SPOT
+        shell: bash
+        run: |
+          set -euo pipefail
+
+          # Files/paths exempt from the SPOT requirement.
+          EXEMPT_REGEX='^(lib/constants\.mjs|test-features\.mjs|ocp|ocp-connect|CHANGELOG\.md|README\.md|docs/|\.github/workflows/alignment\.yml)'
+
+          # Hardcoded port literals to forbid in non-exempt source.
+          FORBIDDEN_PORTS=("3478" "3456")
+
+          FAIL=0
+          for port in "${FORBIDDEN_PORTS[@]}"; do
+            HITS="$(git ls-files | grep -E '\.(mjs|js|ts|json)$' \
+              | xargs grep -n -E "[^0-9]${port}[^0-9]" 2>/dev/null \
+              | grep -v -E "${EXEMPT_REGEX}" \
+              || true)"
+            if [ -n "$HITS" ]; then
+              echo "::error::Hardcoded port literal '${port}' found outside lib/constants.mjs:"
+              echo "$HITS"
+              FAIL=1
+            fi
+          done
+
+          if [ "$FAIL" -ne 0 ]; then
+            cat <<'EOF'
+
+          ============================================================
+          PORT LITERAL SPOT VIOLATION
+          ============================================================
+          A hardcoded TCP port literal was found in a source file
+          that should import from lib/constants.mjs instead.
+
+          Background: this rule exists because between 2026-05-08 and
+          2026-05-13 a stray hardcoded "3478" in scripts/upgrade.mjs
+          and scripts/doctor.mjs cascaded into downstream OpenClaw
+          config writes, taking out the OpenClaw Telegram agent.
+          See v3.16.3 CHANGELOG and lib/constants.mjs header comment.
+
+          Required action:
+            1. Import DEFAULT_PORT (or related constant) from
+               lib/constants.mjs instead of hardcoding the literal.
+            2. If the file genuinely cannot import .mjs (e.g. bash
+               script), add it to EXEMPT_REGEX in this workflow and
+               add a `keep in sync with lib/constants.mjs` comment
+               at the reference.
+            3. For test files that intentionally pin historical ports
+               (test-features.mjs), the regex already exempts them.
+
+          ============================================================
+          EOF
+            exit 1
+          fi
+
+          echo "Port SPOT scan clean."
+
   commit-citation:
     name: commit message citation (soft check)
     runs-on: ubuntu-latest

ALIGNMENT.md

@@ -8,10 +8,14 @@
 
 OCP (Open Claude Proxy) is a **proxy layer** for the Claude Code CLI. It forwards, observes, and multiplexes the traffic that `cli.js` already emits. It is **not** an extension layer. If `cli.js` does not perform a given operation, or performs it differently, OCP does not invent one.
 
+This Core Principle applies in full to **Class A** endpoints (the `cli.js`-mirror surface). A second class of endpoint — **Class B**, the OCP-owned compatibility surface — has its own scope discipline anchored to its own specification authority. See "Scope Clarification: OCP-Owned Compatibility Endpoints (Class B)" below and ADR 0006.
+
 ---
 
 ## Rules
 
+The following Rules apply to **Class A operations** (the `cli.js`-mirror surface — the inbound `/v1/messages` forwarding route, the outbound `/v1/messages` wire call used by `handleUsage()` for rate-limit-header extraction, the OAuth bearer machinery, and any future operations OCP forwards from `cli.js` to Anthropic). For the Class B mapping of each rule, see the Class B section below.
+
 1. **Rule 1 (Grep First).** Before adding, renaming, or changing any endpoint, header, parameter, or response shape, the author must `grep` the reference `cli.js` and record the exact line numbers in the commit message and PR body. An absent grep hit is itself a finding and must be declared.
 
 2. **Rule 2 (No Invention).** OCP must not introduce endpoints, headers, request fields, or response fields that are not present in `cli.js`. Speculative "Claude Code probably uses X" statements are prohibited. If the behavior is not observable in `cli.js`, the feature is out of scope.
@@ -48,6 +52,26 @@ OCP (Open Claude Proxy) is a **proxy layer** for the Claude Code CLI. It forward
 
 The audit pin is updated once per year (see Annual Alignment Audit) and whenever a drift incident forces a re-verification.
 
+### OAuth token-host verification (2026-05-31)
+
+Motivating evidence: the 2026-05-31 code audit (issues #112 / #119 / #123). The OAuth bearer
+machinery is a Class A surface (Rules 1–5). Because `cli.js` now ships as a
+compiled binary, the token-refresh host was re-verified against `claude.exe` (Claude Code
+`2.1.154`) on 2026-05-31 using the compiled-binary protocol — `strings` on the Mach-O, **no
+live OAuth probe** (a `refresh_token` grant would rotate the operator's real credentials):
+
+- **Verified host:** `https://platform.claude.com/v1/oauth/token` — present in the binary
+  byte-for-byte, paired with `OAUTH_CLIENT_ID` in the same `prod` config object (matches
+  `server.mjs` `OAUTH_TOKEN_URL` / `OAUTH_CLIENT_ID`). The legacy `console.anthropic.com/v1/oauth`
+  host is absent (0 hits).
+- **Pinned wrong-host variant:** `console.anthropic.com/v1/oauth/token` is added to the
+  `alignment.yml` blacklist so a future accidental revert to the legacy host hard-fails CI.
+
+The blacklist therefore now holds two kinds of token: (1) known hallucinations (e.g.
+`api.anthropic.com/api/oauth/usage`, the 2026-04-11 drift), and (2) pinned wrong-host variants
+of a *verified* Class A endpoint. A blacklist hit means either a re-introduced hallucination
+**or** a drift to a known-wrong host — both are alignment failures under Rules 2 and 3.
+
 ---
 
 ## Historical Lesson: The 2026-04-11 Drift
@@ -68,20 +92,78 @@ On 2026-04-11, commit `b87992f` ("fix: use dedicated /api/oauth/usage endpoint f
 
 ## Unalignable Policy
 
-A feature is **unalignable** if, after a good-faith search, it cannot be mapped to a specific `cli.js` line range or function.
+A feature is **unalignable** if, after a good-faith search, it cannot be mapped to a specific `cli.js` line range or function (Class A) or to a specific OpenAI specification section AND an authorizing ADR (Class B).
 
 - Unalignable features are **deleted**, not disabled, not feature-flagged, not deprecated.
 - Deletion is the default outcome of an alignment audit finding. The burden of proof is on the feature, not on the auditor.
 - A deletion PR does not require user-facing deprecation notice, because the feature was never legitimately in scope.
-- If a user workflow depended on an unalignable feature, the correct remediation is to upstream the behavior into `cli.js` or to move it out of OCP into a separate tool. OCP does not retain it.
+- If a user workflow depended on an unalignable feature, the correct remediation is to upstream the behavior into `cli.js` (Class A) or into OpenAI's spec (Class B) or to move it out of OCP into a separate tool. OCP does not retain it.
+
+---
+
+## Scope Clarification: OCP-Owned Compatibility Endpoints (Class B)
+
+OCP has two classes of endpoint. Rules 1–5 above were drafted in the aftermath of the 2026-04-11 forwarding drift and are written in the language of a one-to-one proxy; they apply verbatim to **Class A** endpoints. **Class B** endpoints — the OCP-owned compatibility surface where `cli.js` is not the wire authority — have their own scope discipline, anchored to their own specification authority. The full rationale lives in **ADR 0006 (OpenAI Shim Scope)**.
+
+**Class A** — `cli.js`-mirror endpoints. The endpoint exists because `cli.js` performs the equivalent operation and OCP forwards, observes, or multiplexes that operation. Rules 1–5 above apply verbatim. Citation format: `cli.js:NNNN` or `cli.js vE4 <functionName>`.
+
+**Class B** — OCP-owned compatibility endpoints. The endpoint exists because OCP itself surfaces it, with no `cli.js` analogue. Two sub-buckets: **B.1** (OpenAI-compatibility surface — protocol authority is OpenAI's `/v1/chat/completions` specification) and **B.2** (OCP-administrative surface — authority is the ADR that authorized the endpoint's existence).
+
+### Grandfather provision for existing B.2 inventory
+
+ADR 0006 retroactively authorizes the B.2 endpoints listed in the inventory table below, **frozen at their current behaviour as of v3.16.4**. This is a one-time provision; it does not extend to new B.2 endpoints or to B.1 endpoints. Any change to the contract (request shape, response shape, semantics) of a grandfathered B.2 endpoint is treated as a new authorization request and requires either a behaviour-preserving refactor PR or its own ADR. Any new B.2 endpoint, or any new method on a grandfathered B.2 endpoint, requires its own ADR before merge.
+
+### Current Class B inventory
+
+| Endpoint | Method | Sub-bucket | Authorizing ADR |
+|---|---|---|---|
+| `/v1/chat/completions` | POST | B.1 (OpenAI-compat) | ADR 0006 |
+| `/v1/models` | GET | B.1 (OpenAI-compat) | ADR 0006; content sourced from `models.json` per ADR 0003 |
+| `/health` | GET | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/dashboard` | GET | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/sessions` | GET, DELETE | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/logs` | GET | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/status` | GET | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/settings` | GET, PATCH | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/api/keys` | GET, POST | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/api/keys/:id` | DELETE | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/api/keys/:id/quota` | GET, PATCH | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/api/usage` | GET | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/cache/stats` | GET | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/cache` | DELETE | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+
+**Hybrid note.** `/usage` is a hybrid endpoint: the underlying call to `api.anthropic.com/v1/messages` (used to extract `anthropic-ratelimit-unified-*` headers, per the in-file comment block at `server.mjs` line 845–849) is Class A and requires the standard `cli.js` citation; the local synthesis layer that adds `proxy:` stats and `models:` snapshot is Class B and is authorized by ADR 0006. A PR touching only the wire-call layer is Class A; a PR touching only the synthesis layer is Class B; a PR touching both must satisfy both citation requirements.
+
+### Class B citation requirement
+
+Class B PRs cite **the relevant specification section + the authorizing ADR**, in place of `cli.js:NNNN`. Examples:
+
+- B.1: "OpenAI `chat/completions` API, `response_format` parameter (https://platform.openai.com/docs/api-reference/chat/create), authorized by ADR 0006."
+- B.2 (grandfathered): "Authorized by ADR 0006 (grandfathered as of v3.16.4)."
+- B.2 (with its own ADR): "Authorized by ADR 00NN (the ADR that originally authorized the endpoint)."
+
+### Rule mapping for Class B
+
+| Class A rule | Class B mapping |
+|---|---|
+| Rule 1 (Grep First) | Read the cited OpenAI spec section (B.1) or the authorizing ADR (B.2) before writing code. Record the spec URL and ADR number in the PR body. |
+| Rule 2 (No Invention) | OCP must not introduce fields or behaviour not present in OpenAI's spec for the endpoint (B.1) or outside the scope of the authorizing ADR (B.2). For grandfathered B.2 endpoints, "scope" is the v3.16.4 behaviour snapshot. |
+| Rule 3 (Match the Implementation) | Match OpenAI's spec wire-format (B.1) or the ADR's specified behaviour (B.2). |
+| Rule 4 (Unalignable Features Are Deleted) | A Class B endpoint that maps to nothing in OpenAI's spec **and** lacks an authorizing ADR (including not being in the grandfather inventory) is unalignable and is deleted on the same terms as a Class A unalignable feature. |
+| Rule 5 (Cite Line Numbers in Commits) | Cite the OpenAI spec section URL + authorizing ADR number in the commit body (B.1) or the authorizing ADR number alone (B.2). |
+
+### New Class B endpoint procedure
+
+Any new Class B endpoint, or any new method on an existing Class B endpoint (including grandfathered ones), requires its own ADR before merge. An "ADR-less" new Class B endpoint is itself an alignment finding under Rule 4.
 
 ---
 
 ## Annual Alignment Audit
 
 - **Date:** 11 April each year (the anniversary of the `b87992f` drift).
-- **Scope:** Diff the current `cli.js` against the pinned SHA-256 in the Golden Reference section. For every network call in `server.mjs`, re-verify that the corresponding `cli.js` reference still exists at the cited line numbers (adjust citations if line numbers shifted across Claude Code versions).
-- **Output:** A signed audit note committed to `docs/alignment-audits/YYYY-04-11.md`, updating the pin.
+- **Scope (Class A):** Diff the current `cli.js` against the pinned SHA-256 in the Golden Reference section. For every network call in `server.mjs`, re-verify that the corresponding `cli.js` reference still exists at the cited line numbers (adjust citations if line numbers shifted across Claude Code versions).
+- **Scope (Class B):** Audit B.1 endpoints against OpenAI's current `/v1/chat/completions` specification snapshot. Audit B.2 endpoints against their authorizing ADR — for grandfathered endpoints, verify the endpoint behaviour still matches its v3.16.4 snapshot; for ADR-specific endpoints, verify behaviour still matches the ADR. The B.1 specification pin lives in `docs/openai-compat-pin.md` (created alongside the first B.1 audit; not required for ADR 0006 to land).
+- **Output:** A signed audit note committed to `docs/alignment-audits/YYYY-04-11.md`, updating the Class A pin and (once `docs/openai-compat-pin.md` exists) the B.1 pin.
 - **Failure mode:** Any audit finding that cannot be reconciled triggers an immediate deletion PR per the Unalignable Policy.
 
 ---

CHANGELOG.md

@@ -1,6 +1,367 @@
 # Changelog
 
-## v3.15.0 — 2026-XX-XX (release date filled at tag time)
+## v3.24.0 — 2026-07-21
+
+Minor release. Headline: two long-requested **OpenAI-compat features** land — **multimodal vision** (`image_url` parts) and **structured outputs** (`response_format` / JSON schema). Also: the prompt-char budget now derives from the model SPOT instead of a hand-set constant, an agentic-turn bug that dropped the model's final answer is fixed, and `OCP_LOCAL_TOOLS` supports the OpenClaw-backend use case. Four of the six landed from external contributors (@vvlasy-openclaw). Every code PR carried a fresh-context reviewer (Iron Rule 10); no new endpoint, no new `cli.js` wire behavior.
+
+### Added
+
+- **Multimodal vision — OpenAI `image_url` parts (#154, contributed by @vvlasy-openclaw).** `/v1/chat/completions` forwards OpenAI `image_url` content parts to `claude` as native Anthropic image blocks via `--input-format stream-json` (the CLI's own contract — no invented wire shape, verified live). Base64 `data:` URIs by default; remote `http(s)` URLs are off unless `CLAUDE_IMAGE_ALLOW_URL=1` (and even then OCP never fetches them — no SSRF surface). Byte/count caps (`CLAUDE_MAX_IMAGE_BYTES`, `CLAUDE_MAX_IMAGES`, `CLAUDE_MAX_IMAGE_TOTAL_BYTES`), all fail-closed on a misconfigured value. TUI mode returns `400 images_unsupported_in_tui_mode` (it can't carry image blocks) and an image present only in a `system` message returns `400` rather than being silently dropped. README § "Images / Multimodal".
+- **Structured outputs — OpenAI `response_format` (#153, contributed by @vvlasy-openclaw).** `/v1/chat/completions` honors `response_format: { type: "json_schema" | "json_object" }` so OpenAI-SDK clients (Home Assistant AI Tasks, Honcho, scripts) get machine-parseable JSON in `content`. Validates against the schema (incl. `$ref`/`$defs` + `allOf`/`anyOf`/`oneOf` — the shapes the OpenAI SDK emits), retries with a stronger instruction up to `OCP_STRUCTURED_MAX_ATTEMPTS` (default 3, fail-closed), and on exhaustion returns OpenAI's own `refusal` field (200/`content:null`) rather than an invented error. Cyclic-`$ref` schemas fail closed (no stack overflow); a pathologically deep model reply returns a refusal, not a 500 (#181). Single-flight dedup + structured-keyed cache bound the cost. Class B.1 (ADR 0006). README § "Structured Outputs".
+- **SPOT-derived prompt-char budget (#179, ADR 0009).** `MAX_PROMPT_CHARS` default now derives from `max(models.json contextWindow) × 3 chars/token` (600,000 today) instead of the hand-set 150,000 (~37.5k tokens) that silently under-delivered the advertised window ~5×. `CLAUDE_MAX_PROMPT_CHARS` and the settings API remain absolute overrides; a garbage value fails closed to the derived default.
+- **`OCP_LOCAL_TOOLS` — positive local-tools system-prompt wrapper (single-user, loopback only; default off) (#182, contributed by @vvlasy-openclaw).** The `-p` path prepends a wrapper telling the model it has no local filesystem/shell access — correct for a shared gateway, but it makes a personal instance's model (e.g. an OpenClaw agent on its own local OCP) refuse to use the server-side `claude` tools it legitimately has. `=1` swaps in a positive wrapper. Changes **only the prompt**, never the tool surface (`--allowedTools`/`--disallowedTools` untouched; multi-tenant still disallows the FS surface); it does **not** enable client-side `tool_calls` (still unsupported by design). Fail-closed boot gate mirroring `OCP_TUI_FULL_TOOLS` (ADR 0007): refuses to start under `CLAUDE_AUTH_MODE=multi`, a non-loopback bind, or `PROXY_ANONYMOUS_KEY`. Inert (and logged as such) in TUI mode. The active wrapper is folded into the config epoch so toggling it invalidates the standard response cache. No new `cli.js` wire behavior (reuses the already-cited `--system-prompt` flag).
+
+### Fixed
+
+- **Agentic turns dropped the model's final answer (#183, contributed by @vvlasy-openclaw).** On a tool-using turn, `/v1/chat/completions` returned only the opening preamble ("I'll find the repo…") and silently discarded the post-tool-use final answer: aggregate-`assistant` extraction was gated on `isFirstDelta` (which flips false after the first text), and OCP runs pure-aggregate mode (no `--include-partial-messages`), so each of an agentic turn's several assistant messages after the first was lost. Now guards on `sawTextDelta` and accumulates every assistant message (streaming and buffered paths assemble byte-identically).
+- **Deep structured reply returned a 500 instead of a refusal (#181 / #184).** `validateJsonSchema` recurses on the model reply's nesting depth; a ~2000-level-deep reply overflowed the stack → caught `RangeError` → generic 500. A crash-safe façade converts that (only) into a validation miss → refusal; any other throw still surfaces.
+
+## v3.23.0 — 2026-07-17
+
+Minor release. Headline: **the default `sonnet` alias now resolves to Claude Sonnet 5** — a behavior change for every request that omits `model` (pin `claude-sonnet-4-6` by full ID to keep the previous default). Also: Windows-safe upgrade snapshots, two upgrade-system reliability fixes from a live fleet update, the `CLAUDE_SYSTEM_PROMPT` env var made functional, cache-key honesty for config changes, a billing-policy status correction (the 2026-06-15 `-p` split is PAUSED by Anthropic), and a major README restructure. No new endpoint; no new `cli.js` wire behavior. Every code PR carried a fresh-context reviewer (Iron Rule 10).
+
+### Changed
+
+- **Default `sonnet` alias → `claude-sonnet-5` (#168, contributed by @vvlasy-openclaw).** The `sonnet` alias (the model used for every `/v1/chat/completions` request that omits `model`, and OpenClaw's OCP primary via `ocp-connect`) now resolves to `claude-sonnet-5` instead of `claude-sonnet-4-6`. `claude-sonnet-4-6` remains available by full ID for pinning. Mirrors the shipped Claude CLI's own `latest_per_family` mapping (`sonnet → claude-sonnet-5`, verified from binary 2.1.211). Split out from the additive model entry (#152) per Iron Rule 11.
+- **`CLAUDE_SYSTEM_PROMPT` is now functional (#175).** The var was read, documented, and echoed on `/health.systemPrompt` but never reached a request (dead since the `APPEND_SYSTEM_PROMPT` retirement). It is now appended (last, trimmed) to the composed system prompt on the default `-p` path via the new pure `lib/prompt.mjs`; TUI-mode panes are unaffected. Unset ⇒ byte-identical composition to before. README § Environment Variables documents it, including the cache caveat below.
+
+### Fixed
+
+- **Windows-safe upgrade snapshot paths (#167, contributed by @nyxst4ck).** Snapshot directory timestamps now use `-` instead of `:` (Windows forbids `:` in names); legacy colon-named snapshots keep parsing, and `listSnapshots` now orders by **parsed timestamp** (with a deterministic name tie-breaker) so mixed legacy/new names sort chronologically — the initial revision's raw-string sort could delete the newest recovery snapshot at the format boundary and was caught in review; regression tests pin the same-hour mixed-format case.
+- **`ocp update` reliability — two live-incident fixes (#174, closes #173).** (1) The doctor now runs `git fetch --tags` (offline-tolerant) before computing `latest_version` — previously it compared against the locally cached `origin/main`, so machines that hadn't pulled since a release reported "Already at latest" forever. (2) Post-flight now asserts `/health.version` equals the upgrade target (new `postFlightOk` predicate) instead of accepting any `auth.ok` — a stale orphan process holding the port used to pass post-flight while still serving the old version; the failure message now reports the last-seen version and points at `ss -ltnp`/`lsof -i`.
+- **Response-cache key now carries a boot-config epoch (#177, closes #176).** The persistent cache keyed on model+key+params+messages but not on server config that shapes answers (`CLAUDE_SYSTEM_PROMPT`, wrapper text, `CLAUDE_ALLOWED_TOOLS`, `CLAUDE_NO_CONTEXT`) — changing any of these and restarting could serve stale-config answers until TTL expiry. A sha256 config-epoch is folded into every key; any config change is an instant whole-cache invalidation. One-time side effect: existing cache entries miss once after this upgrade.
+
+### Docs
+
+- **Billing-policy status corrected (#171).** Anthropic **paused** the announced 2026-06-15 `claude -p` billing split on its effective date (official help-article citation in README § How It Works): the default `-p` path currently bills the subscription, and TUI-mode is reframed as the ready-made **hedge** for if/when a reworked change lands. All in-force assertions of the split are now date-stamped and conditioned.
+- **LAN mode scoped to chat-class workloads (#171).** New "workload fit" paragraph: multi-device OCP is for text-in/text-out workloads; client-machine coding agents are architecturally out of scope (tools execute on the OCP host).
+- **README restructured, 1205 → ~500 lines (#172).** Operations-manual content moved to `docs/lan-mode.md`, `docs/tui-mode.md`, `docs/troubleshooting.md`, `docs/upgrading.md` (verbatim moves + two canonical dedups; zero content loss verified section-by-section). README keeps the quickstart, the release-kit-pinned reference tables, and summary stubs with links. Plus a staleness sweep (#170): 6-model examples, removal of the never-existed `ocp stop` command, `ocp-connect` claims corrected, current version examples.
+
+## v3.22.1 — 2026-07-17
+
+Minor release: TUI-mode latency and streaming features — **all opt-in and off by default**, so the default request path (`-p` / `--output-format stream-json`) is byte-for-byte unchanged — plus hardening from an independent (Codex) re-review of the streaming work, Windows `claude.exe` startup resolution, and the Claude Sonnet 5 model entry. No new `cli.js` wire behavior and no new endpoint; the new surface is entirely OCP-owned TUI-mode configuration (env vars), startup binary discovery, model metadata, and `/health` observation. Every code PR carried a fresh-context reviewer (Iron Rule 10). (Version note: v3.22.0 was prepared but never tagged; its contents ship here as v3.22.1 together with the additions below.)
+
+### Added
+
+- **Claude Sonnet 5 in the model SPOT (#152, contributed by @vvlasy-openclaw)** — `claude-sonnet-5` added to `models.json` (`contextWindow` 200000 / `maxTokens` 16384 / `reasoning` true, consistent with existing entries), exposed via `/v1/models` and the OpenClaw sync. Purely additive: the `sonnet` alias still resolves to `claude-sonnet-4-6` (the repoint is tracked separately in #168). `ocp-connect`'s model classifier now matches on the model *family* prefix (`claude-sonnet`/`claude-opus`/`claude-haiku`) instead of version-pinned prefixes, so current and future versioned IDs register with correct `reasoning`/`maxTokens` metadata. New referential-integrity tests guard that every alias target exists in `models[]`.
+- **Windows `claude.exe` startup resolution (#161, contributed by @nyxst4ck, diagnosis credit #147 @Justinsato)** — on Windows, `resolveClaude()` now discovers a native `claude.exe` (`%USERPROFILE%\.local\bin`, WinGet Links, WindowsApps, then `where.exe`) and rejects npm `.cmd`/`.bat`/`.ps1` shims, which cannot be spawned without a shell — previously startup resolved a shim and failed. A non-`.exe` `CLAUDE_BIN` on Windows is a fatal error with an actionable hint. The macOS/Linux path is byte-for-byte unchanged. Note: this is startup binary resolution only — full Windows support is not yet claimed (snapshot-path portability is tracked in #167).
+
+### Added — TUI mode (all opt-in, default off)
+
+- **Spawn effort control — `OCP_TUI_EFFORT` (default `low`) (#156)** — the interactive `claude` is now spawned with an explicit `--effort` flag. `low` cuts measured TTFT p50 by ~40% and collapses run-to-run variance ~15× versus an inherited `xhigh`; proxied requests rarely benefit from extended thinking. Set `inherit` to omit the flag and restore the pre-flag HOME-dependent behaviour. Banner-verified to stay on the subscription pool (`· Claude Max`); an invalid value warns and falls back to `low`. README § "Environment Variables".
+- **Warm pane pool — `OCP_TUI_POOL_SIZE` (default `0` / off) (#158)** — pre-boots up to 4 single-use `claude` panes so a request skips the cold boot: measured end-to-end p50 `10.17s` → `6.00s` (−41%) on a Mac mini (Sonnet 4.6, `--effort low`). Opt-in because each warm pane is a live idle process held whether or not a request ever arrives. Panes are single-use (one turn, then killed and replaced in the background), port-scoped (`ocp-tui-<port>-p<hex>`), and coexist with the zombie reaper by a synchronous drain→reap→resume sweep. README §§ "Environment Variables" + "How It Works".
+- **Real SSE streaming — `OCP_TUI_STREAM` (default `0` / off) (#159, #160)** — `stream:true` turns emit real `delta.content` chunks as `claude` generates them, sourced from `claude`'s own `MessageDisplay` hook (registered via `--settings` on the ordinary interactive spawn — banner-verified on the subscription pool). Granularity is block-level, and it moves the *first* byte, not the last. The transcript stays authoritative: streamed text is asserted equal to it at end-of-turn, the auth-banner and truncation gates still run before anything is committed, and a turn whose stream cannot be reconciled is **refused** (SSE error frame, not cached) and counted on `/health` (`tui.streamDivergences`; a silent total-hook-failure is counted separately as `tui.streamZeroDeltaTurns`). Tunables: `OCP_TUI_STREAM_HOLDBACK` (default `100`), `OCP_TUI_STREAM_DIR`, `OCP_TUI_STREAM_POLL_MS`. See ADR 0007 (2026-07-13 amendment). README §§ "Environment Variables" + "How It Works".
+
+### Fixed
+
+- **Streaming auth-banner guard: a null `message_id` on the first hook fire (#160)** — a first `MessageDisplay` fire with a null `message_id` could disarm the auth-banner guard; re-landed after a #159 squash dropped it (`lib/tui/stream.mjs`).
+- **Test suite wrote live, unrevoked API keys into the operator's real key store (#163)** — `npm test` had been opening `~/.ocp/ocp.db` (the running server's DB) and writing two junk `api_keys` rows per run (737 accumulated on the maintainer's host), because the isolation the comments claimed was never wired (ESM import hoisting). `keys.mjs` now honors `OCP_DIR_OVERRIDE` under `NODE_ENV=test` and the suite points at a scratch dir; a child-process probe verifies a production process (no `NODE_ENV`) cannot be redirected.
+- **Streaming holdback floor + billing-pool observation on failed turns (#164)** — (A1) `OCP_TUI_STREAM_HOLDBACK` now clamps up to the safe floor (`100`) with a boot warning, closing a latent auth-banner leak when an operator set a sub-floor value. (A3) the `cc_entrypoint` (billing-pool) observation is now recorded before the honesty gates that throw, so `/health` no longer goes blind to exactly the failed turns most likely to signal a silent degrade to the metered Agent SDK pool.
+- **Test-only key-store redirection vars can no longer reach a server OCP launches (#165)** — (A4) `NODE_ENV`/`OCP_DIR_OVERRIDE` are stripped from every service unit `setup.mjs` writes (`plist-merge`'s `NEVER_PRESERVE`) and from the `ocp restart` manual nohup fallback (`env -u`); #163's overstated "a prod server can NEVER be redirected" comments were softened to name the one residual hand-launch path and the loud `getDb()` "NOT the default" backstop.
+
+### Docs
+
+- **README billing honesty (#162, closes #136)** — removed a feature bullet that promised what the § "honest limits" section forbids.
+- **TUI latency plans + streaming-achievability spike (#155, #157)** — measured latency decomposition, backlog, and the `MessageDisplay`-hook streaming prereq spike under `docs/plans/2026-07-13-tui-latency/`.
+
+## v3.21.1 — 2026-07-07
+
+Patch release: three bug fixes from an independent concurrency/session-lifecycle audit, each its own PR with a fresh-context reviewer (Iron Rule 10). No new `cli.js` wire behavior, no new endpoint, header, or env var; the `/health` field set is unchanged (only value truthfulness improved).
+
+### Fixed
+
+- **TUI session-scope / boot-reap (#148)** — `lib/tui/session.mjs`'s tmux session prefix is now scoped per-instance by listen port (`ocp-tui-<port>-`) instead of a bare host-wide `ocp-tui-` constant, so a second OCP instance on the same host (e.g. a temporary verification instance) can no longer have its live TUI sessions reaped or `kill-server`'d by another instance's boot/periodic sweep. The one-time boot reap also claims exact-shape legacy `ocp-tui-<8hex>` sessions (pre-fix naming) once, to clean up zombies left behind across an in-place upgrade.
+- **`-p` spawn-token mutex + keychain caching (#150)** — the real-HOME token fallback used when the keychain token is within its 5-minute expiry window is now serialized behind a mutex, so concurrent `-p` spawns no longer race the same single-use refresh token against each other (the credential-fork hazard). Added a 30s TTL cache + last-good-label memoization for the keychain read, cutting per-spawn event-loop blocking. The isolation decision (`/health` isolated/real-home reporting) is now re-evaluated per spawn instead of memoized forever, so `/health` no longer misreports a stale decision. New module `lib/spawn-auth.mjs` extracts the pure, unit-testable primitives (mutex, TTL cache, expiry gate, label ordering).
+- **Concurrency queue / disconnect handling (#149)** — the shared semaphore now honors a runtime-lowered `maxConcurrent` immediately (previously a decrease was silently ignored until in-flight tasks finished on their own) and wakes queued waiters right away when the limit is raised. Queued `-p`/TUI requests are now linked to the client's HTTP connection via `AbortSignal`; a client that disconnects while queued is spliced out of the queue instead of still spawning `claude` once a slot frees. A singleflight follower whose leader disconnected now retries instead of inheriting a spurious 500, and a queued-then-disconnected request is no longer recorded as a usage failure or logged as an error (quiet disconnect handling).
+
+## v3.21.0 — 2026-06-25
+
+Cleanup + docs release: TUI dead-code removal, docs honesty, and release prep. No new `cli.js` wire behavior; the default path (`CLAUDE_TUI_MODE` unset) is byte-for-byte unchanged.
+
+### TUI dead-code / footgun cleanup
+
+- **A1 — removed inert entrypoint-env path** (`lib/tui/session.mjs`): deleted `resolveTuiEntrypointEnv()` and the redundant env-strip block in `runTuiTurn`. The `{env}` object passed to `spawnSync` (tmux itself) was the wrong target — tmux does NOT forward the spawning process's environment to the pane; the pane's `claude` gets its env exclusively from the `env` prefix string built inside `buildTuiCmd` (verified live 2026-06-01). The spawnSync env is now intentionally minimal (`HOME` only). Behavior is unchanged: `buildTuiCmd` already handled all claude-specific env vars via its prefix string.
+- **A2 — removed test-only transcript helpers** (`lib/tui/transcript.mjs`): deleted `encodeCwd()` and `transcriptPath()` exports and the tests that pinned them. Production resolves transcripts exclusively via `findTranscriptPath()` (glob by session-id), which is immune to the exact path-encoding rule. No non-test importers existed (grep confirms). A `// TODO` comment near `findTranscriptPath()` notes that a CI fixture-contract test would make claude-schema drift fail loudly.
+- **A3 — removed headless-unusable `--dangerously-skip-permissions` branch** (`lib/tui/session.mjs` + `README.md`): `OCP_TUI_FULL_TOOLS=1` now always takes the `--allowedTools` path. The removed branch pushed `--dangerously-skip-permissions` when `CLAUDE_SKIP_PERMISSIONS=true`; on claude v2.1.x this triggers an interactive bypass-acceptance screen that a headless tmux pane cannot answer → the turn hangs to the wallclock cap and bricks the pane. The working path is `--allowedTools` + scratch-home `settings.json` `additionalDirectories`. `CLAUDE_SKIP_PERMISSIONS` for the `-p` path is unchanged (still used in `server.mjs`).
+
+### Docs
+
+- **Client-tools boundary** (README `§ How It Works`): OCP is a text-prompt bridge only — it does not pass OpenAI `tools`/`functions` or Anthropic `tool_use` blocks to the client. Clients receive assistant TEXT only; client-local tool execution is not supported by design (bypassing `cli.js` = out of scope per `ALIGNMENT.md`).
+- **ToS honesty** (README `§ Deployment model & security`): pooling one Claude subscription across multiple distinct people may violate Anthropic's Consumer ToS and risk account suspension by the abuse classifier. The defensible framing is "one person, your own devices" — friends/team sharing is not. The prior language ("account terms are your call") was accurate but understated the risk.
+- **"Why OCP" posture** (README `§ Why OCP?`): new bullet making explicit that OCP drives the official `claude` CLI as-is — no OAuth token extraction, no binary patching, no protocol invention — so traffic looks like genuine Claude Code (`cc_entrypoint=cli`).
+- **Promotion plan** (`docs/PROMOTION.md`): "stable & visible" strategy covering goal (polish + low-key OSS visibility, NOT growth-hacking given the live ToS/billing risk), pre-requisites (stability first), honest ToS disclosure requirement, items explicitly skipped (multi-backend routing → OLP; gateway model-discovery; raw API passthrough → ALIGNMENT.md scope), TUI toggle as billing-split insurance, and low-key visibility actions. Framed as a recommendation for the maintainer to review, not a committed plan.
+
+### Previously shipped (v3.20.x) — documented here for completeness
+
+- **Default `-p` spawn-home isolation** (v3.20.0 / PR-A): per-request `claude` spawns run in a credential-free minimal scratch HOME (`$HOME/.ocp/spawn-home`, no `.credentials.json`/`settings.json`/plugins) with a neutral cwd and the env token, cutting per-request latency (measured ~10–28s → ~3–7s). Kill-switch: `OCP_SPAWN_REAL_HOME=1`. Active mode shown at startup and on `/health.spawn`.
+- **Bounded concurrency wait-queue** (v3.20.0 / PR-B): excess `-p` requests queue (up to `CLAUDE_MAX_QUEUE`, default 16) instead of being rejected; a full queue returns `HTTP 429` + `Retry-After` (not an opaque 500). New env vars: `CLAUDE_MAX_QUEUE`, `CLAUDE_QUEUE_RETRY_AFTER`. Surfaced on `/health.concurrency` + `/health.stats.queueRejections`.
+- **`ocp restart`** macOS `bootout`+`bootstrap` (v3.20.0 / PR-B): safe restart command that forces launchd to re-read the plist (unlike `kickstart -k` which reuses the cached env).
+- **`/ocp` plugin OpenClaw-2026.5.27 compat** (v3.20.0 / PR-C): gateway plugin updated for the current OpenClaw API version.
+
+## v3.20.1 — 2026-06-13
+
+TUI-mode auth hardening: fixes the recurring `Please run /login · API Error: 401` (the PI231 incident) and reaps leaked defunct `claude` sessions. ([#141](https://github.com/dtzp555-max/ocp/pull/141))
+
+### Fixed
+
+- **TUI 401 / credential corruption (#141)** — interactive `claude` prefers `~/.claude/.credentials.json` over the `CLAUDE_CODE_OAUTH_TOKEN` env var (unlike `-p` mode, where the env token wins). OCP TUI's per-request spawn + `kill-session` cycle raced claude's single-use refresh-token rotation, corrupting the refresh token to an empty string → permanent 401 that `claude /login` couldn't fix (each new spawn re-corrupted it). This bit Linux/file-based hosts specifically (macOS reads credentials from the Keychain, so Mac mini was immune). **Fix:** when `CLAUDE_CODE_OAUTH_TOKEN` is set, the TUI claude now runs in a credential-free scratch HOME (`<HOME>/.ocp-tui/home`, overridable by `OCP_TUI_HOME`) seeded with onboarding + cwd-trust but **no `.credentials.json`**, so the env token is the only credential and claude never runs the refresh path. Recurrence-proof — a later `claude login` can no longer break TUI. Also: `buildTuiCmd` passes `CLAUDE_CODE_OAUTH_TOKEN` to the spawn, and `reapStaleTuiSessions` reaps defunct `claude` sessions (tmux-server-owned zombies) via `kill-server` when no foreign session remains, plus a 15-min idle-gated periodic reap. When the env token is unset, behaviour is byte-for-byte unchanged (real-home + credentials.json). Two independent fresh-context reviewers (Iron Rule 10) + a live PI231 portability test (works with a corrupt credentials.json present). Authorized by the ADR 0007 PR-D amendment (Class B).
+
+### Environment variables
+
+- `CLAUDE_CODE_OAUTH_TOKEN` — when set on a TUI host, TUI authenticates via this long-lived token in a credential-isolated home (recommended; immune to credentials.json corruption).
+- `OCP_TUI_HOME` — overrides the TUI scratch home; if you previously pointed it at your real home, unset it to get the credential-isolated default.
+
+## v3.20.0 — 2026-06-10
+
+TUI-mode billing-safety hardening for the 2026-06-15 Anthropic billing split. A 5-dimension multi-agent audit (adversarial verification + live tests on all three hosts — PI231 / Oracle / Mac mini, claude 2.1.104 / 2.1.114 / 2.1.170) found the TUI subscription-pool path could silently bill the metered Agent SDK pool or poison the cache under realistic failure modes. Three PRs, each with a fresh-context reviewer (Iron Rule 10) and CI; the default path (`CLAUDE_TUI_MODE` unset) is byte-for-byte unchanged.
+
+### TUI — honesty & cache correctness (#137)
+
+- **C-1** — `callClaudeTui` now throws on a claude-CLI auth-failure banner (e.g. `Please run /login · API Error: 401 …`, `Failed to authenticate. API Error: 401 …`) instead of returning it as a real answer, so it is never cached, singleflight-shared, or counted as a model success. Conservative detector (whole trimmed text ≤100 chars + `API Error: 4xx` + auth keyword + no code/quote char); overridable via `CLAUDE_TUI_ERROR_PATTERNS`. Live-reproduced on PI231.
+- **C-2** — `readTuiTranscript` distinguishes a complete turn from a wallclock-truncated partial (`truncated` flag); `callClaudeTui` throws `tui_wallclock_truncated` so a partial is never cached or counted as success.
+- **C-3** — `verifyEntrypoint` reads the `entrypoint` field from any transcript line, not just `{system, turn_duration}` — some claude builds emit zero turn_duration lines (live-confirmed on Oracle's claude 2.1.114), which previously left the billing-drift assertion blind on those builds.
+- **C-4 (paste)** — short prompts (e.g. `hi`) could never pass paste-landing detection; threshold lowered. Live-reproduced on PI231.
+
+### TUI — concurrency & observability (#139)
+
+- **Concurrency** — `OCP_TUI_MAX_CONCURRENT` (default 2) bounds concurrent interactive `claude` boots via a queuing semaphore (`lib/tui/semaphore.mjs`); the slot is released on throw so honesty-gate / spawn failures never leak it; bounded wait-queue → `tui_queue_full` (503). Independent of the global `MAX_CONCURRENT` (8) — a TUI turn is a heavy per-request cold-boot of tmux+claude + up to 120s wallclock.
+- **Observability** — additive `/health` `tui` block (`enabled` / `entrypointMode` / `lastEntrypoint` / `entrypointMismatches` / `inflight` / `maxConcurrent`) so an operator can poll for a silent `sdk-cli` metered-pool drift (the audit's top risk) instead of grepping journald. Authorized by the ADR 0007 PR-B amendment under the ALIGNMENT grandfather provision (additive, behaviour-preserving — every pre-existing `/health` field unchanged).
+
+### Operations (#138)
+
+- `docs/runbooks/615-canary.md` — the 2026-06-15 credit-balance canary: quiesce, read the Agent SDK credit balance (manual — no programmatic API exists for that pool; OCP's `/usage` headers are subscription rate-limit data, not the credit pool), one TUI canary turn, confirm `entrypoint:cli` in the transcript, green/red decision tree, periodic auto-mode self-classification mini-canary.
+- `docs/runbooks/tui-flip-rollback.md` — flip/rollback per deployment (systemd `daemon-reload`; launchd `bootout`/`bootstrap`, not `kickstart -k`).
+- `setup.mjs` auth quick-test gated behind `OCP_SKIP_AUTH_TEST=1` (the `claude -p` probe draws from the metered Agent SDK pool after 6/15).
+
+### New environment variables
+
+- `OCP_TUI_MAX_CONCURRENT` — max concurrent interactive TUI turns (default 2) (#139).
+- `OCP_SKIP_AUTH_TEST` — skip the `claude -p` auth probe in `setup.mjs` (default off) (#138).
+
+## v3.19.0 — 2026-06-02
+
+TUI-mode reliability + proxy-purity release. Two fixes diagnosed and verified live on both test hosts (PI231 / Oracle, claude 2.1.104 / 2.1.114), each its own PR with a fresh-context reviewer (Iron Rule 10), then an adversarial multi-host test battery (0 hangs / 0 crashes / 0 injection / 0 leaks). The default path (`CLAUDE_TUI_MODE` unset) is byte-for-byte unchanged.
+
+### TUI
+
+- **#130** — Fixed the "stuck typing" hang on large multi-line prompts. Three root causes: (1) terminal-turn detection only recognized `{system, turn_duration}`, which older claude builds (e.g. 2.1.114) don't emit → the reader ran to the wallclock and returned partial text; now also accepts an `assistant` line with a final `stop_reason` (`end_turn`/`stop_sequence`/`max_tokens`), while `tool_use` stays non-terminal. (2) Large prompts pasted via `send-keys -l` delivered embedded newlines as separate Enter events → the prompt never landed; now uses `tmux load-buffer` + `paste-buffer -p` (bracketed paste, atomic). (3) The paste-landed check false-positived on claude's empty curly-quote placeholder → Enter fired into an empty box; now positive-signal-only (`[Pasted text]` / prompt text) with a readiness/paste-verify poll + fast-fail (deterministic ~5s error instead of a 120s wallclock hang).
+- **#4** — TUI-mode never injects the host's `CLAUDE.md` / auto-memory into proxied turns. OCP is a proxy: the proxied client (OpenClaw / an IDE) owns its own context and memory. `buildTuiCmd` now always sets `CLAUDE_CODE_DISABLE_CLAUDE_MDS` + `CLAUDE_CODE_DISABLE_AUTO_MEMORY` (unconditional — proxy purity is not an opt-in). Verified live with a marker `CLAUDE.md`: obeyed by the proxied turn before the fix, blocked after, on both hosts. Residual host-context vectors (managed-policy / `settings.json` / output-styles) tracked in #133. The env is delivered via an `env`-prefix on the tmux pane command (tmux does not forward the spawning process's environment, and `new-session -e` requires tmux ≥3.2 while the cloud host runs 2.7).
+
+## v3.18.0 — 2026-06-01
+
+Hardening release from a multi-agent code audit (1 P0 + 14 P2 + 2 P3 findings, each adversarially verified and independently reviewed) plus three follow-ups (#123–#125). Every change shipped as its own PR with a fresh-context reviewer (Iron Rule 10). The single-user default path (`AUTH_MODE=none`, no TUI) is behavior-identical **except** the `/health` change in #109.
+
+### Security
+
+- **#109 (P0)** — `/health` no longer advertises `PROXY_ANONYMOUS_KEY` to remote callers by default. The `anonymousKey` field is gated behind a new `PROXY_ADVERTISE_ANON_KEY=1` opt-in env var; localhost callers are always exempt. Prevents any LAN-reachable device from harvesting a working, quota-spending bearer credential from the unauthenticated `/health` endpoint. **Behavior change:** `ocp-connect` zero-config Path A now requires the server to set `PROXY_ADVERTISE_ANON_KEY=1`; otherwise pass `--key` or use anonymous access.
+- **#114** — Dashboard escapes all DB-sourced strings (key names, usage rows) before `innerHTML`; the revoke button uses a `data-` attribute + listener instead of an inline `onclick` a quote could break out of; `POST /api/keys` validates key names server-side (`[A-Za-z0-9 ._-]{1,64}`).
+- **#124** — Dashboard status/plan summary cards escaped too (uniform defense-in-depth over all `innerHTML` sinks).
+- **#111** — Streaming error paths strip filesystem paths from claude error text / stderr before sending them to clients (`sanitizeError`), matching the non-streaming path.
+
+### Reliability / correctness
+
+- **#110** — Non-array `messages` is rejected with a 400 (was silently hanging the connection until socket timeout); OpenAI array `content` is flattened into the prompt instead of dumped as raw JSON; a streamed upstream error now emits an SSE `error` frame instead of a success-looking `finish_reason:"stop"`.
+- **#111** — `res.on("close")` escalates SIGTERM→SIGKILL on client disconnect (closes a narrow re-occurrence of the #37 concurrency-slot leak on the hottest exit path); `overallTimer` is cleared on semantic completion so a slow-exiting child can't record a spurious post-success timeout; per-key quota is documented as best-effort (bounded overshoot ≤ `MAX_CONCURRENT`, cache hits uncounted).
+- **#113** — CLI/installer hardening: `ocp-plugin` restart uses the live uid + `dev.ocp.proxy`/`ocp-proxy` labels and drops the unsafe `pkill` fallback; `ocp-connect` quotes + `chmod 600`s the persisted key; `setup.mjs` XML-escapes and newline-validates injected service-unit secrets.
+
+### Alignment / governance
+
+- **#112** — OAuth token-refresh host (`platform.claude.com/v1/oauth/token`) re-verified against the compiled cli.js v2.1.154 (`strings`, no live probe) and recorded in `ALIGNMENT.md`; usage-probe and default request model now derive from `models.json` (ADR 0003 SPOT) instead of hardcoded IDs.
+- **#123** — The legacy `console.anthropic.com/v1/oauth/token` host is pinned in the `alignment.yml` blacklist so a future OAuth-host drift hard-fails CI; the blacklist now documents its dual purpose (known hallucinations + pinned wrong-host variants of a verified Class A endpoint).
+
+### TUI
+
+- **#115** — The TUI LAN gate refuses any non-loopback bind (not just literal `0.0.0.0`); the achieved `cc_entrypoint` is asserted each turn and a `tui_entrypoint_mismatch` warning is logged on a silent degrade to the metered sdk-cli pool.
+
+### Refactor
+
+- **#125** — `isLoopbackBind` extracted to `lib/net.mjs`, shared by `server.mjs` and the test suite (was duplicated via a copy-paste mirror).
+
+### New environment variables
+
+- `PROXY_ADVERTISE_ANON_KEY` — opt-in (default off); advertise `PROXY_ANONYMOUS_KEY` on the public `/health` body for remote zero-config discovery (#109).
+
+## v3.17.1 — 2026-05-31
+
+### Fix — code-audit P1/P2 hardening
+
+Fixes from a multi-agent code audit (3 P1 + 5 P2, adversarially verified). The single-user default path (`AUTH_MODE=none`, no TUI) is behavior-identical.
+
+**Availability / correctness (P1):**
+- Guard `proc.stdin` against EPIPE — a fast-failing spawned `claude` (auth error, bad model, large prompt) no longer crashes the single-process daemon.
+- Add `unhandledRejection`/`uncaughtException`/`clientError` safety nets + wrap all request-body read loops — a client aborting mid-upload no longer crashes the daemon.
+- TUI transcript reader: only `turn_duration` is terminal (was also `tool_use`), which silently truncated any TUI turn that used a built-in tool.
+
+**Security gates / cache integrity (P2):**
+- `AUTH_MODE=multi`: the default spawn now passes `--disallowedTools` (Bash/Read/Write/Edit/…) so a guest prompt cannot drive operator-filesystem tools. Single-user path unchanged.
+- `/sessions` (DELETE), `/settings` (PATCH), `/logs`, `/usage`, `/status` are now admin-gated (were dispatched before the admin check).
+- Streaming path no longer caches an `is_error` response as success (cache-poisoning fix).
+- TUI fail-loud guard extended to `none`+`0.0.0.0` (unless `OCP_TUI_ALLOW_LAN=1`) and `+ PROXY_ANONYMOUS_KEY`.
+- TUI `send-keys` paste uses `-l` (literal) so a prompt equal to a tmux key token (e.g. `C-c`) is typed, not interpreted.
+
+---
+
+## v3.17.0 — 2026-05-31
+
+### Provider — default claude invocation ported to stream-json + `--system-prompt` (Phase 6c)
+
+OCP's default (non-TUI) claude spawn moves from `claude -p --output-format text` to `claude --output-format stream-json --verbose --no-session-persistence --system-prompt <wrapper>` (no `-p`). The NDJSON event stream is parsed into the assembled response. Benefits: ~64% per-request cost reduction and anti-hallucination via `--system-prompt` tool-use suppression. Clients see no API change — the OpenAI-compatible request/response shapes are identical. Faithful port of OLP's production-verified implementation; covered by 17 new stream-json parser tests.
+
+⚠️ **Billing note:** from 2026-06-15 this default path carries `cc_entrypoint=sdk-cli` and bills against the Agent SDK credit pool. Use the new opt-in `CLAUDE_TUI_MODE` (below) to keep traffic on the Pro/Max subscription pool.
+
+---
+
+### feat(tui): opt-in CLAUDE_TUI_MODE — serve via interactive claude (cc_entrypoint=cli / subscription pool), single-user only; default stream-json path unchanged
+
+From 2026-06-15 Anthropic routes `claude -p` / `--output-format` invocations to the Agent SDK credit pool (`cc_entrypoint=sdk-cli`). This feature adds an opt-in bridge: when `CLAUDE_TUI_MODE=true`, OCP serves each request via a real interactive `claude` session (no `-p`, no `--output-format`) so it carries `cc_entrypoint=cli` and bills against the Pro/Max subscription.
+
+The complete string response is read from claude's native JSONL session transcript and replayed to callers as a normal OpenAI completion or chunked SSE. Clients see no API change. The default stream-json path is byte-for-byte unchanged when `CLAUDE_TUI_MODE` is unset.
+
+**Security:** single-user / single-operator only. Never enable on a multi-user OCP. See ADR 0007 and README § "Subscription-pool (TUI) mode".
+
+New env vars: `CLAUDE_TUI_MODE`, `CLAUDE_TUI_WALLCLOCK_MS`, `OCP_TUI_CWD`, `OCP_TUI_HOME`.
+New ADR: `docs/adr/0007-tui-interactive-mode.md`.
+New modules: `lib/tui/transcript.mjs`, `lib/tui/session.mjs` (shipped in preceding commits on this branch).
+
+---
+
+### Model — add claude-opus-4-8
+
+Add `claude-opus-4-8` as the newest Opus to `models.json` (index 0, newest first). Repoint `aliases.opus` from `claude-opus-4-7` to `claude-opus-4-8`. `claude-opus-4-7` remains in the list callable by literal id. `legacyAliases.claude-opus-4` left pointing at `claude-opus-4-7` (no change — legacy alias tracks the prior generation). README Available Models table and model-count references updated accordingly.
+
+---
+
+## v3.16.4 — 2026-05-13
+
+### Refactor — port-literal SPOT + CI guardrail
+
+Closes the structural side of the port-drift cascade addressed by v3.16.2
+and v3.16.3. Those two releases reverted plist / plugin / scripts back to
+3456 line-by-line, but the underlying invitation to drift — a hardcoded
+port literal scattered across six source files — was still intact.
+
+Changes:
+
+- **New `lib/constants.mjs`** — single source of truth for shared literals.
+  Exports `DEFAULT_PORT = 3456`, `LOCAL_HOST = "127.0.0.1"`,
+  `OPENAI_API_BASE = "/v1"`, `LOCAL_PROXY_URL`.
+- **`server.mjs:127`, `setup.mjs:36`, `scripts/upgrade.mjs:137`,
+  `scripts/doctor.mjs:84` + `:205`, `scripts/sync-openclaw.mjs:73`** —
+  all replaced with imports from `lib/constants.mjs`. Behavior is
+  identical; the literal `3456` now exists in exactly one place per
+  language (`lib/constants.mjs` for `.mjs`, `ocp` + `ocp-connect` for
+  bash, `test-features.mjs` for pinned historical-port tests).
+- **`.github/workflows/alignment.yml`** — extended the path filter to
+  `setup.mjs`, `scripts/**`, `lib/**`, `ocp`, `ocp-connect`. Added a new
+  `port-spot` hard-fail job that greps for any hardcoded `3478` or `3456`
+  literal in `.mjs/.js/.ts/.json` outside the EXEMPT_REGEX (which lists
+  `lib/constants.mjs`, `test-features.mjs`, the bash CLIs, docs, and the
+  workflow itself). Any future PR re-introducing a hardcoded port
+  literal will be blocked at CI before it can cascade.
+- Doc comments in `server.mjs` env-var summary and `setup.mjs` usage
+  banner reworded so the literal `3456` no longer appears as
+  documentation text (CI grep is intentionally aggressive — it does not
+  parse comments — so doc strings reference `DEFAULT_PORT from
+  lib/constants.mjs` instead).
+
+No behavior change for any user. `CLAUDE_PROXY_PORT` env var remains
+the runtime override; the only difference is the unset-env fallback
+now flows through one shared constant.
+
+ALIGNMENT.md hard-requirements: this PR modifies `server.mjs` (one-line
+import + one literal swap, mechanical). No cli.js operation changed;
+the citation requirement does not apply. SPOT principle (Rule 2 spirit)
+is the entire motivation.
+
+## v3.16.3 — 2026-05-13
+
+### Fixes — completes v3.16.2 port-drift revert
+
+v3.16.2 reverted the plugin / `openclaw.plugin.json` / README / Mac mini
+plist back to `3456` (the historical source default since `593d0dc`), but
+missed three places in `scripts/` that still defaulted to `3478`. Those
+three lines were the residual cascade source: every time `ocp doctor` or
+`ocp upgrade` ran without `CLAUDE_PROXY_PORT` in the env, they probed
+`3478`, reported "OCP not responding" against a healthy 3456 instance,
+and (in the case of OpenClaw sync follow-ups on the maintainer's host)
+re-introduced 3478 into downstream config.
+
+Changes:
+
+- `scripts/upgrade.mjs:137` — default port `3478` → `3456`.
+- `scripts/doctor.mjs:84` — default port `3478` → `3456`.
+- `scripts/doctor.mjs:205` — default port `3478` → `3456`.
+
+No behavior change for users who set `CLAUDE_PROXY_PORT` explicitly; env
+still takes precedence. The fix only affects the unset-env fallback,
+which now matches `server.mjs:126` and the rest of the codebase.
+
+Test plan: existing `test-features.mjs` cases that pin
+`CLAUDE_PROXY_PORT=3478` continue to pass — they use the env path, not
+the default.
+
+## v3.16.2 — 2026-05-12
+
+### Fixes — corrects v3.16.1
+
+The v3.16.1 fix was directionally correct (plugin now reads env first, falls back to a hardcoded default) but **the narrative and the hardcoded default were both wrong**.
+
+What v3.16.1 said: "OCP server moved to 3478 default in v3.14+; plugin lagged at 3456."
+What is actually true:
+- **OCP server source default has been `3456` since `593d0dc` (initial release) and has never changed.** Every line in `server.mjs`, `setup.mjs`, and the `ocp` CLI still uses `3456` as the documented and code-level default.
+- The single OCP installation observed on `3478` is the maintainer's Mac mini, whose plist was rewritten with `--port 3478` during a PR #71 dogfood smoke-test accident on 2026-05-08 (see `~/.cc-rules/memory/learnings/subagent_setup_mjs_prod_host_collision.md`). The plist drift was never reconciled back to source default, and v3.16.1 incorrectly canonised the post-accident value as if it had been a release decision.
+
+This release:
+- Restores the plugin fallback to `http://127.0.0.1:3456` to match server source default.
+- Updates `openclaw.plugin.json` `configSchema.proxyUrl.default` back to `3456`.
+- Restores README §"Environment Variables" `CLAUDE_PROXY_PORT` default to `3456`.
+- Plugin reads `OCP_PROXY_URL` env (full URL) first, then `CLAUDE_PROXY_PORT` env (port only), then falls back to `3456`. Hosts whose OCP plist injects a non-default port must also inject the same `CLAUDE_PROXY_PORT` into the OpenClaw plist for the plugin to follow.
+- Maintainer's Mac mini plist was reverted from `3478` to `3456` as part of this release deploy (no source change reflects this; it was a one-host correction).
+
+### Governance
+
+- No `cli.js` citation needed (no `server.mjs` change). ALIGNMENT.md Rule 2 not engaged.
+
+## v3.16.1 — 2026-05-12 (superseded — narrative incorrect; see v3.16.2 erratum)
+
+### Fixes (as shipped — note erratum above)
+
+- **OCP plugin port lag** — `ocp-plugin/index.js` hard-coded `http://127.0.0.1:3456`. ~~While OCP server moved to 3478 in v3.14+,~~ **(corrected v3.16.2: no such move ever happened.)** The Mac mini's plist was on `3478` only as residue from a dogfood accident. Result: `/ocp` slash commands from the home Telegram bot returned "OCP error: fetch failed". v3.16.1 changed the plugin default to `3478` (wrong direction; v3.16.2 reverts to `3456`).
+
+### Governance
+
+- No `cli.js` citation needed (no `server.mjs` change). ALIGNMENT.md Rule 2 not engaged.
+
+## v3.16.0 — 2026-05-10
+
+### Features
+
+- **`ocp doctor --check oauth`** (PR #93) — fast path that runs only the OAuth check, skipping
+  version detection / from-version / git operations / models endpoint. ~50ms vs. full doctor's
+  ~200-500ms. Use cases: AI agent repair loops, post-`claude auth login` verify, quick health
+  gates. Help text in `cmd_doctor_help` now reflects working behaviour.
+- **`ocp update --rollback --gc`** — manually garbage-collect old upgrade snapshots.
+  Retention policy: keep last 5 snapshots OR snapshots newer than 30 days OR the single most
+  recent (always-keep safety net). `--dry-run` previews. Successful `ocp update` runs auto-GC
+  at the end of the full path; light path does not (no snapshot created there).
+
+### Behavior changes
+
+- After a successful cross-minor `ocp update`, the auto-GC emits `[gc] removed N old snapshots`
+  to stderr if any were collected. Safe to ignore; manual gc is `ocp update --rollback --gc`.
+
+### Governance
+
+- No `cli.js` citation needed (no `server.mjs` change). ALIGNMENT.md Rule 2 not engaged.
+- PR #93 (--check oauth) merged separately; this release bundles it with the GC feature.
+
+## v3.15.1 — 2026-05-10
+
+### Fixes
+
+- **doctor: dynamic `latest_version` from `origin/main:package.json`** — v3.15.0 doctor used a hard-coded `latest = "v3.14.0"` fallback, which made any v3.15.0+ install report `kind = upgrade` (against a stale value). `ocp update` would then attempt `git checkout v3.14.0` — a downgrade. Doctor now fetches `git -C ~/ocp show origin/main:package.json` to determine the actual latest version; on failure (offline, fresh clone with no remote), falls back to `currentVersion` so `kind = noop` instead of recommending a downgrade.
+
+## v3.15.0 — 2026-05-10
 
 ### Features
 

dashboard.html

@@ -132,6 +132,10 @@ function fmtChars(n) {
   return n > 1000 ? (n/1000).toFixed(0) + "K" : String(n);
 }
 
+function escapeHtml(s) {
+  return String(s ?? "").replace(/[&<>"']/g, c => ({ "&": "&amp;", "<": "&lt;", ">": "&gt;", '"': "&quot;", "'": "&#39;" }[c]));
+}
+
 function barColor(pct) {
   if (pct >= 80) return "bar-red";
   if (pct >= 50) return "bar-amber";
@@ -144,8 +148,8 @@ async function refreshStatus() {
   const r = data.requests || {};
 
   document.getElementById("status-cards").innerHTML = `
-    <div class="card"><div class="label">Status</div><div class="value"><span class="tag ${p.status === 'ok' ? 'tag-ok' : 'tag-err'}">${p.status || '?'}</span></div><div class="sub">v${p.version || '?'}</div></div>
-    <div class="card"><div class="label">Uptime</div><div class="value">${p.uptime || '?'}</div></div>
+    <div class="card"><div class="label">Status</div><div class="value"><span class="tag ${p.status === 'ok' ? 'tag-ok' : 'tag-err'}">${escapeHtml(p.status || '?')}</span></div><div class="sub">v${escapeHtml(p.version || '?')}</div></div>
+    <div class="card"><div class="label">Uptime</div><div class="value">${escapeHtml(p.uptime || '?')}</div></div>
     <div class="card"><div class="label">Requests</div><div class="value">${r.total || 0}</div><div class="sub">${r.active || 0} active</div></div>
     <div class="card"><div class="label">Errors</div><div class="value">${r.errors || 0}</div><div class="sub">${r.timeouts || 0} timeouts</div></div>
     <div class="card"><div class="label">Sessions</div><div class="value">${p.activeSessions || 0}</div></div>
@@ -160,15 +164,15 @@ async function refreshStatus() {
     document.getElementById("plan-cards").innerHTML = `
       <div class="card">
         <div class="label">Session (5h)</div>
-        <div class="value">${s.percent || '?'}</div>
+        <div class="value">${escapeHtml(s.percent || '?')}</div>
         <div class="bar-bg"><div class="bar-fill ${barColor(sPct)}" style="width:${sPct}%"></div></div>
-        <div class="sub">Resets in ${s.resetsIn || '?'}</div>
+        <div class="sub">Resets in ${escapeHtml(s.resetsIn || '?')}</div>
       </div>
       <div class="card">
         <div class="label">Weekly (7d)</div>
-        <div class="value">${w.percent || '?'}</div>
+        <div class="value">${escapeHtml(w.percent || '?')}</div>
         <div class="bar-bg"><div class="bar-fill ${barColor(wPct)}" style="width:${wPct}%"></div></div>
-        <div class="sub">Resets in ${w.resetsIn || '?'}</div>
+        <div class="sub">Resets in ${escapeHtml(w.resetsIn || '?')}</div>
       </div>
     `;
   }
@@ -181,21 +185,21 @@ async function refreshUsage() {
     const tbody = document.querySelector("#key-usage-table tbody");
     tbody.innerHTML = (data.byKey || []).map(k => `
       <tr>
-        <td>${k.key_name}</td>
+        <td>${escapeHtml(k.key_name)}</td>
         <td>${k.requests}</td>
         <td>${k.successes}</td>
         <td>${k.errors}</td>
         <td>${fmtTime(k.avg_elapsed_ms)}</td>
-        <td class="mono">${k.last_request || '-'}</td>
+        <td class="mono">${escapeHtml(k.last_request || '-')}</td>
       </tr>
     `).join("") || '<tr><td colspan="6" style="color:#475569">No usage data yet</td></tr>';
 
     const rtbody = document.querySelector("#recent-table tbody");
     rtbody.innerHTML = (data.recent || []).slice(0, 20).map(r => `
       <tr>
-        <td class="mono">${r.created_at?.slice(11, 19) || '?'}</td>
-        <td>${r.key_name}</td>
-        <td>${r.model}</td>
+        <td class="mono">${escapeHtml(r.created_at?.slice(11, 19) || '?')}</td>
+        <td>${escapeHtml(r.key_name)}</td>
+        <td>${escapeHtml(r.model)}</td>
         <td>${fmtChars(r.prompt_chars)}</td>
         <td>${fmtChars(r.response_chars)}</td>
         <td>${fmtTime(r.elapsed_ms)}</td>
@@ -216,13 +220,16 @@ async function refreshKeys() {
     const tbody = document.querySelector("#keys-table tbody");
     tbody.innerHTML = (data.keys || []).map(k => `
       <tr>
-        <td>${k.name}</td>
-        <td class="mono">${k.keyPreview}</td>
-        <td class="mono">${k.created_at}</td>
+        <td>${escapeHtml(k.name)}</td>
+        <td class="mono">${escapeHtml(k.keyPreview)}</td>
+        <td class="mono">${escapeHtml(k.created_at)}</td>
         <td><span class="tag ${k.revoked ? 'tag-err' : 'tag-ok'}">${k.revoked ? 'revoked' : 'active'}</span></td>
-        <td>${k.revoked ? '' : `<button class="btn btn-sm btn-danger" onclick="revokeKeyUI('${k.name}')">Revoke</button>`}</td>
+        <td>${k.revoked ? '' : `<button class="btn btn-sm btn-danger" data-revoke="${escapeHtml(k.name)}">Revoke</button>`}</td>
       </tr>
     `).join("");
+    tbody.querySelectorAll("button[data-revoke]").forEach(btn =>
+      btn.addEventListener("click", () => revokeKeyUI(btn.getAttribute("data-revoke")))
+    );
   } catch(e) { /* not admin */ }
 }
 

docs/PROMOTION.md

@@ -0,0 +1,100 @@
+# OCP Promotion Strategy — "Stable & Visible"
+
+> **This document is a recommendation for the maintainer to review and adjust, not a committed plan.**
+> It reflects the project's current posture (post-v3.21.0) and should be revisited whenever
+> the Anthropic billing / ToS environment changes significantly.
+
+---
+
+## 1. Goal: Polish + Low-Key OSS Visibility
+
+The goal is **stability and quiet discoverability**, not growth-hacking. OCP is a personal power tool
+that has been open-sourced because others can benefit from it. The right audience finds it via GitHub
+search, issue threads in related projects, and word of mouth — not viral posts.
+
+**Explicitly avoid:**
+
+- HN / Reddit front-page pushes, influencer outreach, or any campaign that would attract a large
+  influx of users before the ToS/billing situation has settled. Anthropic is actively tightening
+  billing and enforcement on subscription-sharing (the June-15 Agent-SDK billing split is
+  *paused*, not cancelled — and consumer-ToS enforcement on multi-person sharing is a live risk).
+  A high-traffic spotlight right now would draw scrutiny that a low-profile project avoids.
+- Promising features that require bypassing the `claude` CLI (raw API calls, OAuth extraction, etc.)
+  — that would violate `ALIGNMENT.md` and the ToS simultaneously.
+
+---
+
+## 2. Pre-Requisite: Stability First
+
+Do not promote until the house is in order:
+
+- [x] The concurrency / latency perf fixes are shipped (v3.20.x–v3.21.0).
+- [x] Docs honesty is complete (client-tools boundary, ToS sharing disclosure, this doc).
+- [ ] The June-15 Agent-SDK billing split is either confirmed cancelled or OCP has a confirmed
+      stable path (TUI toggle as insurance — see §5 below).
+
+Promoting a project that has known rough edges in docs or stability only generates support burden
+and negative first impressions.
+
+---
+
+## 3. Honest ToS Disclosure on Sharing
+
+Any promotion materials must carry the same disclosure as `README.md § "Deployment model & security"`:
+
+> Pooling a single Claude subscription across **multiple distinct people** may violate Anthropic's
+> Consumer Terms of Service and risk account suspension. The defensible framing is "one person,
+> your own devices". Friends/team sharing is not.
+
+This framing should appear in any README badge, linked blog post, or issue comment that mentions
+LAN sharing. It is not a disclaimer that discourages usage — it is honest positioning that protects
+both the project and its users.
+
+---
+
+## 4. What to Explicitly Skip
+
+These items are **not gaps in OCP** — they are deliberate stance decisions:
+
+- **Multi-backend routing** (routing to OpenAI, Gemini, Llama, etc.) — that is the sibling [OLP
+  project](https://github.com/dtzp555-max/olp)'s role. OCP stays Claude-only by design.
+- **Gateway model-discovery** (auto-detecting which models a remote server offers) — not needed
+  for OCP's single-provider, single-subscription model. `models.json` is the SPOT.
+- **Raw Anthropic API passthrough** (bypassing the `claude` CLI) — out of scope per `ALIGNMENT.md`.
+
+Do not add these to OCP roadmaps or respond to feature requests for them with "planned" — the
+correct answer is "that's OLP territory" or "out of scope per ALIGNMENT.md".
+
+---
+
+## 5. TUI Toggle as Insurance
+
+The `CLAUDE_TUI_MODE` opt-in is the primary mitigation if the June-15 billing split reactivates
+and makes the default `-p` path draw from the metered Agent SDK credit pool.
+
+Keep the TUI toggle:
+- Functional and tested across the three deployment hosts.
+- Documented in the README, including the security constraints (single-user only).
+- Easily discoverable for users who get unexpectedly metered.
+
+If the split reactivates, the recommended operator path is: set `CLAUDE_TUI_MODE=true` +
+`CLAUDE_CODE_OAUTH_TOKEN` → credential-isolated scratch home → subscription pool. That path is
+already shipped and documented.
+
+---
+
+## 6. Low-Key Visibility Actions (when §2 pre-requisites are met)
+
+- Keep the GitHub README polished and honest — it is the primary landing page.
+- Respond promptly to issues and PRs — the project's reputation is built on reliability, not
+  marketing.
+- Add OCP to the `awesome-claude` / `awesome-llm-tools` lists if they exist and allow self-PRs
+  — low-effort, targeted, reaches the right audience.
+- When related projects (Cline, OpenCode, OpenClaw, Continue.dev) post about local Claude proxies,
+  a short factual comment linking to OCP is appropriate — not spam.
+- Maintain the `CHANGELOG.md` with clear, honest summaries — users who are already running OCP
+  are the best vector for word-of-mouth.
+
+---
+
+*Last updated: v3.21.0 cleanup cycle. Maintainer should re-read before any external promotion.*

docs/adr/0006-openai-shim-scope.md

@@ -0,0 +1,132 @@
+# 0006 — OpenAI Shim Scope: Class A vs Class B Endpoints
+
+- **Date**: 2026-05-20
+- **Status**: Proposed — owner reviewing
+- **Authors**: project maintainer (with AI drafting assistance)
+- **Related**: `ALIGNMENT.md` (the constitution); ADR 0002 (Alignment Constitution provenance, PR #20, commit 2853088); PR #99 by external contributor (triggering incident — OpenAI `response_format` honoring on `/v1/chat/completions`)
+
+## Context
+
+`ALIGNMENT.md` was drafted in the aftermath of the 2026-04-11 drift (commit `b87992f` — fabricated `/api/oauth/usage` endpoint) and ratified in PR #20 / commit 2853088. Its five Rules are written in the language of a one-to-one proxy: Rule 1 (Grep First), Rule 2 (No Invention), Rule 3 (Match the Implementation), Rule 4 (Unalignable Features Are Deleted), Rule 5 (Cite Line Numbers in Commits). All five anchor explicitly on `cli.js` as the golden reference. This is correct and binding for the endpoints OCP was originally designed to forward — `/v1/messages`, `/api/oauth/*`, and the rate-limit-header extraction path that backs `/usage` — because for those, `cli.js` is the literal wire authority and any deviation is a drift risk.
+
+OCP also exposes a second class of endpoint that the constitution does not currently distinguish: **OpenAI-compatible surface** that exists so non-Claude-Code clients (Honcho, OpenWebUI, OpenAI SDK consumers, BYO scripts) can talk to claude via OCP. The flagship is `/v1/chat/completions`, which translates between OpenAI's request/response schema and `cli.js`'s native protocol. `cli.js` never speaks OpenAI's wire format — by construction it cannot, because OpenAI and Anthropic are different vendors with different protocols. There is no `cli.js:NNNN` to cite for OpenAI's `messages[].role` field handling, OpenAI's streaming `delta` shape, OpenAI's `stop` event names, or OpenAI's `response_format` parameter. The protocol authority for these is OpenAI's published specification, not `cli.js`.
+
+The structural gap surfaced when PR #99 (external contributor `jaekwon-park`) added support for the OpenAI `response_format` request field on `/v1/chat/completions`. A strict reading of Rule 2 ("OCP must not introduce request fields that are not present in `cli.js`") blocks the PR. But the same strict reading also blocks the existence of `/v1/chat/completions` itself — every OpenAI-shaped field on that endpoint is, by definition, not in `cli.js`. The endpoint has been in OCP since before the constitution was written and is used by real downstream consumers. The constitution and the endpoint cannot both be correct under the current reading.
+
+The 2026-04-11 drift remains the cautionary tale that drove the constitution and remains binding. The drift was not "OCP exposed an endpoint that wasn't in `cli.js`" — it was specifically "OCP claimed to forward `cli.js`'s `/api/oauth/usage` call when no such call exists in `cli.js`." That is a Class A failure mode: a forwarding endpoint that lied about what it was forwarding. The fix to that failure mode (Rules 1, 2, 3, 5; CI blacklist; reviewer gate) was correct then and is correct now. This ADR does not relitigate that decision and does not soften Rules 1–5 for the class of endpoint they were designed to discipline.
+
+What this ADR does is acknowledge that OCP has two classes of endpoint, and that the discipline that fits Class A does not fit Class B without distortion. Class B needs its own anchor (OpenAI's specification) and its own authorization gate (an ADR per endpoint), so contributors know exactly which rule set applies to their PR and so Class B never becomes a backdoor for "OCP can do anything OpenAI-shaped."
+
+## Decision
+
+Introduce an explicit two-class taxonomy of OCP endpoints:
+
+- **Class A — `cli.js`-mirror endpoints.** Endpoints that exist because `cli.js` performs the equivalent operation and OCP forwards, observes, or multiplexes that operation. Rules 1–5 of `ALIGNMENT.md` apply verbatim. The citation requirement is `cli.js:NNNN` (or `cli.js vE4 <functionName>`).
+
+- **Class B — OCP-owned compatibility endpoints.** Endpoints that exist because OCP itself surfaces them, with no `cli.js` analogue. They fall into two sub-buckets:
+  - **B.1 — OpenAI-compatibility surface.** Endpoints implementing OpenAI's published API contract so non-Anthropic clients can use OCP. The protocol authority is OpenAI's specification.
+  - **B.2 — OCP-administrative surface.** Endpoints that exist purely to operate the proxy itself (health, dashboard, key management, cache control). The authority for these is the ADR that authorized the endpoint's existence.
+
+For Class B endpoints, the citation requirement shifts from `cli.js:NNNN` to **(a)** the relevant specification section (OpenAI spec section for B.1, or the authorizing ADR for B.2) **and (b)** the ADR that authorized the endpoint's existence in the first place.
+
+### Grandfather provision for existing B.2 inventory
+
+ADR 0006 retroactively authorizes the existing B.2 endpoints listed in the inventory table below, **frozen at their current behaviour as of v3.16.4**. This is a one-time grandfather provision intended to avoid a 12-ADR back-fill burden for endpoints that have existed in OCP since before any constitutional governance was written.
+
+The grandfather provision is narrowly scoped:
+
+- It covers only the B.2 endpoints enumerated in the inventory table as of this ADR's merge date.
+- It freezes those endpoints at their **current behaviour**. Any change to the request shape, response shape, or semantics of a grandfathered B.2 endpoint is treated as a new authorization request and requires either (a) a behaviour-preserving refactor PR with no contract change, or (b) its own ADR.
+- It does **not** authorize new B.2 endpoints. Any new B.2 endpoint, or any new method on a grandfathered B.2 endpoint, requires its own ADR before merge.
+- It does **not** extend to B.1 (OpenAI-compat) endpoints. B.1 endpoints are bounded by OpenAI's published specification, not by a behaviour snapshot — there is no grandfather equivalent for them.
+
+The structural intent is: take the one-time hit of declaring "current B.2 surface is authorized" cleanly, then make every future addition pay the ADR-per-endpoint cost. This prevents Class B from becoming a backdoor for general OCP-owned-surface invention while not blocking the present ADR on twelve back-fill PRs.
+
+### Current Class B inventory (enumerated from `server.mjs`)
+
+The following endpoints exist today in `server.mjs` and are Class B (no `cli.js` analogue):
+
+| Endpoint | Method | Sub-bucket | Authorizing ADR |
+|---|---|---|---|
+| `/v1/chat/completions` | POST | B.1 (OpenAI-compat) | ADR 0006 |
+| `/v1/models` | GET | B.1 (OpenAI-compat) | ADR 0006; content sourced from `models.json` per ADR 0003 |
+| `/health` | GET | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/dashboard` | GET | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/sessions` | GET, DELETE | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/logs` | GET | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/status` | GET | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/settings` | GET, PATCH | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/api/keys` | GET, POST | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/api/keys/:id` | DELETE | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/api/keys/:id/quota` | GET, PATCH | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/api/usage` | GET | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/cache/stats` | GET | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+| `/cache` | DELETE | B.2 (administrative) | ADR 0006 (grandfathered as of v3.16.4) |
+
+For Class A reference, the current Class A inventory is `/v1/messages` (forwarded directly to `api.anthropic.com/v1/messages`) and the OAuth bearer / rate-limit-header machinery used by `handleUsage()` (which calls `https://api.anthropic.com/v1/messages` to extract `anthropic-ratelimit-unified-*` headers, per the in-file comment at line 845–849). The `GET /usage` endpoint surface itself is Class B (administrative augmentation: it adds `proxy:` and `models:` blocks not present in any upstream API), but the data fetch underlying it is Class A — see "Hybrid endpoints" below.
+
+### Hybrid endpoints
+
+`/usage` is a hybrid: the wire call out to `api.anthropic.com/v1/messages` is Class A and must continue to cite `cli.js`; the local synthesis on top (`proxy:` stats block, `models:` snapshot, response shape) is Class B and is authorized by this ADR. Any future change strictly to the wire-call layer is Class A; any change strictly to the synthesis layer is Class B. A PR touching both must satisfy both citation requirements.
+
+## What does NOT change
+
+The following continue to apply verbatim and are not weakened by this ADR:
+
+- **Rules 1, 2, 3, 4, 5 of `ALIGNMENT.md`** for all Class A endpoints. The 2026-04-11 drift discipline is unchanged. Class A PRs still require `cli.js:NNNN` citations, still must match `cli.js`'s wire format byte-for-byte, and still face the Unalignable Policy if the citation cannot be produced.
+- **CI blacklist** (`.github/workflows/alignment.yml`). The known-hallucinated token list (currently `api.anthropic.com/api/oauth/usage`) continues to be greppable-and-failable on every PR.
+- **Reviewer gate** (CLAUDE.md hard requirements + Iron Rule 10). Implementation author may not self-approve; a fresh-context reviewer opens `cli.js` at the cited lines for Class A PRs.
+- **Annual Alignment Audit** on 11 April. The Class A audit (re-verify each `server.mjs` Class A reference against the pinned `cli.js` SHA-256) continues unchanged.
+- **Unalignable Policy.** A Class A endpoint that cannot be traced to a `cli.js` reference is still deleted, not deprecated.
+- **Historical Lesson section in `ALIGNMENT.md`.** The 2026-04-11 drift remains the named cautionary incident, with commit SHAs intact.
+
+## What additionally applies to Class B
+
+The following are new and apply only to Class B endpoints:
+
+1. **OpenAI specification as protocol authority (B.1).** The OpenAI compatibility surface follows OpenAI's published `/v1/chat/completions` specification (https://platform.openai.com/docs/api-reference/chat/create) — not OCP imagination, not "OpenAI probably does X," not generalization from adjacent OpenAI endpoints. The same anti-invention discipline that Rule 2 imposes for `cli.js` applies, with OpenAI's spec substituted as the reference.
+
+2. **ADR-authorized endpoint existence.** Any new Class B endpoint, or any new Class B endpoint method, requires its own ADR before merge. The grandfather provision above covers existing B.2 inventory only. An "ADR-less" Class B endpoint added after this ADR merges is itself an alignment finding and is subject to deletion under a Class B equivalent of the Unalignable Policy (see Rule 4 mapping in `ALIGNMENT.md`'s new section).
+
+3. **Class B citation format.** Class B PRs cite (a) the relevant specification section and (b) the authorizing ADR. Example for B.1: "OpenAI `chat/completions` API, `response_format` parameter (https://platform.openai.com/docs/api-reference/chat/create), authorized by ADR 0006." Example for B.2: "Authorized by ADR 0006 (grandfathered)" for grandfathered endpoints, or "Authorized by ADR 00NN" for endpoints with their own ADR.
+
+4. **Class B audit cadence.** Class B endpoints are audited annually alongside the Class A audit. B.1 endpoints are audited against OpenAI's current `/v1/chat/completions` specification snapshot. B.2 endpoints (grandfathered or ADR-specific) are audited against their authorizing ADR — for grandfathered endpoints, the audit verifies the endpoint behaviour still matches its v3.16.4 snapshot; for ADR-specific endpoints, the audit verifies behaviour still matches the ADR. The B.1 specification pin lives in `docs/openai-compat-pin.md` (to be created alongside the first B.1 audit; not a prerequisite for this ADR to land).
+
+5. **Reviewer expectation.** The fresh-context reviewer for a Class B PR opens the cited OpenAI spec section (B.1) or the authorizing ADR (B.2) instead of opening `cli.js`. The "I am not the commit author" rule and the "explicit approval comment naming the verified reference" rule continue.
+
+## Consequences
+
+**Positive**
+
+- PR #99 becomes mergeable with a one-line scope declaration ("Class B — extends `/v1/chat/completions` per ADR 0006") plus the existing alignment-evidence section adapted to Class B citation format. The structural ambiguity that blocked it is removed.
+- Future Class B contributors have a clear template and a defensible scope: "extend `/v1/chat/completions` for an OpenAI-spec field that's already in OpenAI's spec" is a well-formed PR; "add a new OCP-invented field that looks OpenAI-shaped" is not, and the same anti-invention discipline that protects Class A protects Class B.
+- The Class A surface is structurally unchanged. Reviewers reading the new `ALIGNMENT.md` see a clean Class A regime with all five Rules intact, plus an enumerated and explicitly scoped Class B carve-out.
+- The administrative endpoint surface (B.2) is no longer in a "is this even allowed under the constitution?" limbo. The grandfather provision cleanly authorizes the current inventory; new B.2 endpoints must earn their ADR.
+
+**Negative**
+
+- OCP now maintains a second alignment surface. OpenAI's `/v1/chat/completions` specification is also a moving target (OpenAI ships changes more than once per year, including breaking ones), so the B.1 audit has real work attached.
+- The grandfather provision freezes the current B.2 behaviour. If a grandfathered B.2 endpoint has a latent bug or undesirable behaviour, "fixing" it is a contract change and now requires an ADR (or a behaviour-preserving refactor). This is intentional friction to prevent silent contract drift.
+- Contributors must now choose Class A or Class B on every PR. Some will misclassify. The PR template's required Class A/B radio (see PR template update) and the reviewer's spec-or-cli verification step are the structural counter-measures.
+
+**Mitigations**
+
+- The Class B inventory is small (currently 14 endpoints) and is enumerated explicitly in `ALIGNMENT.md`. New entries require an ADR per item 2 above, so the inventory cannot grow silently.
+- Anthropic-side change frequency (which drives Class A audit cost) is structurally higher than OpenAI's `chat/completions` shape, which has been stable across multiple OpenAI API versions. The marginal B.1 audit cost is low. The grandfathered B.2 audit cost is also low — most of those endpoints have not changed in months.
+- The B.1 specification pin in `docs/openai-compat-pin.md` lets the audit anchor on a specific OpenAI spec snapshot, the same way the Class A pin anchors on a specific `cli.js` SHA-256. Drift detection then works the same way for both classes.
+
+## Historical Lesson — explicit non-relitigation
+
+This ADR does not relitigate the 2026-04-11 drift. The drift commit `b87992f` was Class A — it claimed `cli.js` forwarded a call that `cli.js` did not in fact make. The fix (constitution + CI blacklist + reviewer gate) was correct and remains binding for Class A. This ADR carves out Class B because the discipline that fits Class A does not fit a class of endpoint where `cli.js` is not the wire authority — not because the discipline was wrong, and not because the drift lesson is any less load-bearing.
+
+A reviewer or future maintainer reading this ADR should not infer: "OCP relaxed its alignment rules." The Class A regime is structurally identical to the version that shipped in PR #20. What changed is that the constitution now names the scope of that regime precisely (the class of endpoint for which `cli.js` is the wire authority) instead of implicitly applying it to every endpoint, including ones the regime was never designed for.
+
+## Alternatives considered
+
+**(a) Refuse Class B as a category; close PR #99 and delete `/v1/chat/completions`.** This would resolve the structural ambiguity by enforcing Rule 2 maximally — if `cli.js` doesn't speak OpenAI's protocol, neither does OCP. Rejected: there is an existing user base on the OpenAI-compat surface, the surface is genuinely useful (it is OCP's bridge to non-Claude-Code agents), and deletion would be a load-bearing user-facing breakage in service of a doctrinal point that the constitution was never designed to make. The constitution was a response to the 2026-04-11 forwarding drift, not a charter against any OCP-owned surface.
+
+**(b) Soften Rule 2 to "OCP must not introduce surface area not present in `cli.js` OR not authorized by an ADR."** This is the obvious diff and would unblock PR #99 with the smallest possible textual change. Rejected because it loses the precision that Class A needs. A combined Rule 2 means a Class A reviewer has to read the PR description twice to figure out which authority applies. The Class A/B split makes the question explicit at the PR-template level (the author picks the class) and at the reviewer level (the reviewer opens the appropriate reference). The cost of the split is one new section in `ALIGNMENT.md`; the benefit is no ambiguity in either class.
+
+**(c) Move `/v1/chat/completions` and all OpenAI-compat surface out of OCP into a separate "ocp-openai-shim" repository.** This would cleanly resolve the scope question by moving Class B out of OCP entirely. Rejected as premature: the maintainer is one person, the OpenAI-compat surface today is a single endpoint plus its support, and the operational cost of two repositories (separate releases, separate CI, separate version coordination) exceeds the cost of one constitution with two named classes. If the OpenAI-compat surface ever grows to the size where a separate repo is justified, ADR 0006 is the natural pivot point — at that future date, the carve-out becomes a separation.
+
+**(d) Twelve-ADR back-fill for the existing B.2 inventory before this ADR can merge.** Considered and rejected on cost grounds. Each back-fill ADR would be a short paragraph explaining what an existing endpoint does and why it's allowed; the educational value is low and the merge friction is high (12 PRs through the reviewer gate). The grandfather provision above achieves the same authorization outcome in one paragraph, while still requiring an ADR for any future B.2 endpoint. The trade-off: grandfathered endpoints are not individually documented to ADR-depth. Mitigation: the inventory table in `ALIGNMENT.md` lists every grandfathered endpoint by path and method, so the audit surface remains explicit.

docs/adr/0007-tui-interactive-mode.md

@@ -0,0 +1,393 @@
+# ADR 0007 — TUI Interactive Mode (subscription-pool bridge)
+
+**Date:** 2026-05-31
+**Status:** Accepted — amended by PR-4 (entrypoint hardening), PR-B (observability + concurrency), PR-C (env-token auth + defunct-reaping), PR-D (credential-isolated home — corrects PR-C)
+**Deciders:** project maintainer
+**Authority:** claude CLI v2.1.158 interactive mode — verified live on the test host that sessions launched without `-p` / `--output-format` carry `cc_entrypoint=cli` (subscription pool), not `cc_entrypoint=sdk-cli` (Agent SDK credit pool). Mechanism verified on cli.js v2.1.104; live-confirmed on v2.1.158.
+
+---
+
+## Context
+
+On 2026-05-14 Anthropic announced (effective 2026-06-15) a billing split that routes requests by `cc_entrypoint`:
+
+| `cc_entrypoint` value | Billing pool |
+|-----------------------|-------------|
+| `cli` | Pro/Max subscription pool |
+| `sdk-cli` | Agent SDK credit pool (~$20/mo on Pro = easily exhausted) |
+
+OCP's existing path (`claude --output-format stream-json -p`) sets `cc_entrypoint=sdk-cli`. After 2026-06-15 every OCP request will draw from the Agent SDK pool rather than the subscription.
+
+The structural response: add an opt-in mode that drives a real **interactive** `claude` session (no `-p`, no `--output-format`), which carries `cc_entrypoint=cli` and therefore bills against the subscription. The response text is read from claude's native JSONL transcript instead of from `stdout`.
+
+This is a personal-use A-path feature (single-user, single-subscription host). It is **not** a multi-tenant isolation layer.
+
+### Source-verified entrypoint mechanism (PR-4 amendment)
+
+Claude CLI's `main()` calls a startup function (`t$A` in the compiled bundle) that sets
+`process.env.CLAUDE_CODE_ENTRYPOINT` **only if unset** to:
+
+```
+(argv has -p/--print/--init-only/--sdk-url  OR  !process.stdout.isTTY) ? "sdk-cli" : "cli"
+```
+
+The billing header reads `cc_entrypoint = process.env.CLAUDE_CODE_ENTRYPOINT ?? "unknown"`.
+The `"unknown"` branch is dead code for any real `main()` spawn — the startup function always
+sets a value on unset env. The **real risk** is not `"unknown"`: it is a **lost TTY** (e.g. stdout
+redirected or a non-PTY spawn) silently flipping the self-classification to `"sdk-cli"` and
+drawing from the metered pool.
+
+`cc_entrypoint` is one of ~6 upstream run-mode signals. The **dominant discriminator** is the
+system-prompt identity block ("official CLI" vs "Claude Agent SDK"), which is driven by genuine
+interactivity (no `-p`, no `--output-format`, real PTY) and is overridable by no env var. This
+is the real reason the tmux/no-`-p` approach works: the spawn is genuinely interactive, not just
+labelled as such.
+
+---
+
+## Decision
+
+Add `CLAUDE_TUI_MODE=true` as an opt-in flag in `server.mjs`.
+
+### How it works
+
+1. Each request spawns a fresh tmux session running `claude --model <M> --session-id <UUID> --strict-mcp-config --disallowedTools 'mcp__*'` (no `-p`, no `--output-format`).
+2. The spawn result is checked immediately: if `tmux new-session` returns a non-zero exit status (or a falsy result), the request is aborted with `tui_spawn_failed: tmux session not created` **before** the boot sleep. This is the spawn/PTY gate — OCP must not issue a billing request without a verified interactive session.
+3. The serialized prompt (from `messagesToPrompt`) is pasted via `tmux send-keys … "$(cat file)"` + a separate `Enter` key event.
+4. The answer is read from claude's native JSONL transcript at `<HOME>/.claude/projects/<encoded-cwd>/<session-id>.jsonl`, polling until a `turn_duration` system event or the wall-clock cap (`CLAUDE_TUI_WALLCLOCK_MS`, default 120 s).
+5. The string answer is returned to OCP's existing downstream (singleflight → cache write-back → `completionResponse` / `streamStringAsSSE`) — **same contract as `callClaude`**.
+6. Streaming requests are buffered then replayed as chunked SSE (no real token streaming — deliberate; "don't build fragile features"). **Superseded for `stream:true` when `OCP_TUI_STREAM=1` — see the 2026-07-13 amendment below. The buffered path remains the default and is unchanged.**
+
+### Billing-classifier labeling (`OCP_TUI_ENTRYPOINT`, PR-4)
+
+`CLAUDE_CODE_ENTRYPOINT` on the spawn env is managed by `resolveTuiEntrypointEnv(env, mode)`
+(exported from `lib/tui/session.mjs`, pure, testable). The function **always deletes any
+inherited value first** so a stray env var from OCP's own parent process can never leak in and
+mislabel the billing header. Then:
+
+| `OCP_TUI_ENTRYPOINT` | Behaviour |
+|----------------------|-----------|
+| `cli` (default) | Sets `CLAUDE_CODE_ENTRYPOINT=cli` deterministically — subscription-pool classification. **Honest only because the spawn is a genuine interactive PTY** (tmux pane, no `-p`, stdout not redirected, `new-session` verified). |
+| `auto` | Deletes the key → claude self-classifies via `t$A` (TTY → `cli`). Use to observe/diagnose the real TTY-derived value. |
+| `off` | Leaves the env exactly as inherited — diagnostics / honesty audit only. |
+
+**Governing rule (verbatim):** *OCP may make a true value deterministic; it may never assert a
+value the spawn's real state contradicts. When it cannot make the claim true (e.g. cannot
+guarantee a PTY), it fails/drops the request — it does not force the signal.*
+
+This is why the spawn/PTY gate (step 2 above) is load-bearing for `mode="cli"`: if `new-session`
+fails, there is no PTY, so asserting `cli` would be dishonest. Abort rather than lie.
+
+OCP never suppresses the billing header (anti-fingerprinting: we do not mask the spawn).
+
+### 2026-06-15 verification protocol
+
+Run one quiesced canary request in TUI-mode and watch the **Agent SDK credit balance** (not the
+request header). If the balance drops, the subscription pool is unreachable via spawn. Per the
+constitution (`ALIGNMENT.md`), the response is to **drop the Anthropic provider** rather than
+escalate spoofing.
+
+Version caveat: mechanism verified on cli.js v2.1.104 + live on v2.1.158. Re-verify after any
+major cli.js upgrade.
+
+### Default behaviour is unchanged
+
+When `CLAUDE_TUI_MODE` is unset (the default), no code path touches `callClaudeTui` or `runTuiTurn`. `upstreamCall === callClaude` and streaming uses `callClaudeStreaming` — byte-for-byte identical to the pre-TUI code path.
+
+### Kill-switch
+
+Unset `CLAUDE_TUI_MODE` (or set it to any value other than `"true"`) → stream-json path restored immediately on next restart.
+
+### Home strategy
+
+> **Superseded by the PR-D amendment below for the env-token case.** As of PR-D, `TUI_HOME`
+> is computed by `resolveTuiHome()`: when `CLAUDE_CODE_OAUTH_TOKEN` is set (and `OCP_TUI_HOME`
+> is unset) the default is a **credential-free scratch home**, not the real home. The
+> descriptions below remain accurate for the **no-env-token** case and the **explicit
+> `OCP_TUI_HOME` override** case.
+
+- **Real-home (default when NO env token, `OCP_TUI_HOME` unset):** claude runs with the operator's own `~/.claude/` — shared credentials, existing onboarding, no OAuth fork risk. `ensureTuiCwdTrusted` seeds the trust record for the scratch cwd in the real `~/.claude.json` (atomic write).
+- **Scratch-home opt-in (`OCP_TUI_HOME=<path>`, no env token):** a dedicated `HOME` that symlinks `~/.claude/.credentials.json` from the real home (token is never copied) and seeds a stripped `~/.claude.json` (no project history, trusts only the scratch cwd). **Caveat:** claude rewrites `.credentials.json` on OAuth token refresh, replacing the symlink with a regular file — this forks the credentials. Use this legacy symlink mode only with a dedicated OAuth or for ephemeral testing. (The PR-D env-token mode avoids this caveat entirely — no credentials file to fork.)
+
+### Working directory
+
+`TUI_CWD = OCP_TUI_CWD || $HOME/.ocp-tui/work` (dedicated scratch cwd). Transcripts land under `<HOME>/.claude/projects/<encoded-cwd>/` — a stable, single location separate from the operator's real project histories. The directory is created automatically on first request.
+
+### MCP hard-disable
+
+`--strict-mcp-config` (no `--mcp-config` argument) prevents account-attached managed MCP servers from connecting. Belt-and-braces: `--disallowedTools 'mcp__*'` blocks any MCP tool invocation even if a server were somehow loaded. Built-in tools (Bash, Read, etc.) are left enabled on the A-path (single-user, acceptable).
+
+### Session namespace
+
+All tmux sessions use the prefix `ocp-tui-`. The prefix-scoped reaper (`reapStaleTuiSessions`) kills only `ocp-tui-*` sessions, never `olp-tui-*` or any other prefix. A stale-session cleanup runs once at OCP boot when `TUI_MODE` is on.
+
+---
+
+## SECURITY — PROMINENT WARNING
+
+**TUI-mode is SINGLE-USER / SINGLE-OPERATOR ONLY.**
+
+`claude` runs as the OCP process owner with full filesystem access regardless of `HOME` setting. Home selection is **not** user isolation. If OCP is serving multiple users or guest API keys:
+
+- A guest prompt would run `claude` with the **operator's** filesystem access.
+- An adversarial prompt could exfiltrate files, run shell commands, or exhaust the subscription.
+
+**Never enable `CLAUDE_TUI_MODE=true` on an OCP instance that serves untrusted callers or multiple users.**
+
+The B-path (multi-tenant isolation) requires:
+1. `--tools ""` (no built-in tools)
+2. Per-key ephemeral `HOME` (isolated credentials + no cross-key project pollution)
+3. Sandbox runtime (e.g. `@anthropic-ai/sandbox-runtime`)
+
+B-path is **deferred** and is not implemented in this ADR. Until B-path lands, TUI-mode must only be enabled on a personal single-user OCP.
+
+---
+
+## Observability and concurrency (PR-B amendment)
+
+**Date:** 2026-06-10
+**Status:** Accepted — amends ADR 0007.
+**Motivation:** the post-PR-A code audit, findings C-4 (P1) and C-5 (P1).
+
+### C-4 — independent concurrency bound for the TUI path
+
+The global `MAX_CONCURRENT` gate lives in `spawnClaudeProcess()` (the `-p` / stream-json
+path). `callClaudeTui()` never calls `spawnClaudeProcess` — it calls `runTuiTurn()`, which
+cold-boots a full interactive `claude` inside a fresh tmux session. So the TUI path had **no**
+concurrency bound: N concurrent TUI requests spawned N simultaneous cold-boot tmux+claude
+processes. On a small host (e.g. a Pi 4 serving a family) a burst of ~5 is an OOM risk and
+also multiplies subscription rate-limit pressure.
+
+PR-B adds an **independent** limiter for the TUI path (`lib/tui/semaphore.mjs`,
+`TuiSemaphore`):
+
+- **`OCP_TUI_MAX_CONCURRENT`, default `2`.** Rationale: a TUI turn is heavy — a per-request
+  cold-boot of tmux+claude plus up to `CLAUDE_TUI_WALLCLOCK_MS` (120 s) of wallclock — so a
+  small host cannot run many at once. `2` is the conservative default that keeps a Pi-class
+  host alive under a family burst while still allowing some overlap. It is deliberately **not**
+  the same knob as `MAX_CONCURRENT` (default 8): the two pools have different shapes (a
+  stream-json spawn is cheap and fast; a TUI turn is a heavy cold-boot + long wallclock), so
+  coupling them would mis-size one of the two paths.
+- **Queue, don't reject.** The limiter **queues** (awaits a slot), mirroring the spirit of
+  `MAX_CONCURRENT` — requests are not dropped on contention. To bound memory against a runaway
+  client, the wait queue itself is capped (`maxQueue`, default 32× the limit); when the queue
+  is full `run()` rejects with `tui_queue_full`, surfaced as a 503 — deterministic backpressure
+  rather than silent OOM.
+- **Slot released in a `finally`.** `TuiSemaphore.run(fn)` releases the slot in a `finally`, so
+  any throw — PR-A's honesty gates (`tui_wallclock_truncated`, `tui_upstream_error`), a
+  `tui_paste_not_landed`, or a `tui_spawn_failed` — can never leak a slot.
+
+This limiter has **zero effect when `TUI_MODE` is off**: `callClaudeTui` is never reached, so
+the semaphore is never entered. The default stream-json path is untouched.
+
+### C-5 — operator-visible drift surface on `/health` (additive)
+
+The `tui_entrypoint_mismatch` warning only reached journald. After the 2026-06-15 flip, a
+silent `sdk-cli` drift (the documented top risk in this ADR — a lost TTY flipping the
+self-classification to the metered Agent SDK pool) would drain metered credits **invisibly**.
+PR-B adds a `tui` block to the `/health` JSON response so an operator can poll it:
+
+```
+tui: {
+  enabled:              <TUI_MODE>,
+  entrypointMode:       <OCP_TUI_ENTRYPOINT>,   // cli | auto | off
+  lastEntrypoint:       <last observed cc_entrypoint, e.g. "cli", or null>,
+  entrypointMismatches: <count of cli-expected-but-got-other turns>,
+  inflight:             <current concurrent TUI turns>,
+  queued:               <turns waiting for a slot>,
+  maxConcurrent:        <OCP_TUI_MAX_CONCURRENT>
+}
+```
+
+`lastEntrypoint` is recorded and `entrypointMismatches` incremented inside `callClaudeTui` in
+the same mismatch branch that already emits the journald warning (via `recordTuiEntrypoint`).
+`inflight` / `queued` / `maxConcurrent` come from the C-4 semaphore. When `TUI_MODE` is off the
+block still appears with `enabled:false` (cheap, harmless) so the response shape is stable for
+consumers regardless of mode.
+
+### ALIGNMENT authorization for the `/health` change
+
+`/health` is a **grandfathered B.2 endpoint** under ADR 0006, frozen at its v3.16.4 behaviour.
+`ALIGNMENT.md`'s grandfather provision states: *"Any change to the contract (request shape,
+response shape, semantics) of a grandfathered B.2 endpoint is treated as a new authorization
+request and requires either a behaviour-preserving refactor PR or its own ADR."*
+
+This amendment **is** that authorization. The argument:
+
+- The change is **additive**: it adds one new top-level field (`tui`) containing only new
+  sub-fields. **No existing `/health` field is changed, renamed, removed, or re-typed**, and no
+  existing semantics change. Existing `/health` consumers (the dashboard, `ocp-connect`,
+  monitoring) read the fields they already read and are unaffected — the change is
+  **behaviour-preserving** for them, which is exactly the bar the grandfather provision sets for
+  a non-ADR contract change.
+- The TUI observability surface is an **intrinsic part of the TUI feature** whose authorizing
+  authority is **this ADR (0007)**, not a brand-new B.2 endpoint. We are not adding a new B.2
+  endpoint or a new method (which would each require their own fresh ADR under the New Class B
+  endpoint procedure) — we are extending the response of an existing grandfathered endpoint with
+  fields that report state owned by an ADR-0007 feature. ADR 0007 is the natural home for that
+  authority, and this amendment records it explicitly.
+- `cli.js` does not perform this operation — `/health` is OCP-owned (Class B), so no `cli.js`
+  citation applies; the citation is this ADR + ADR 0006 (grandfathered B.2) per
+  `ALIGNMENT.md`'s Class B citation requirement.
+
+### `OCP_TUI_MAX_CONCURRENT` summary
+
+| Env var | Default | Meaning |
+|---|---|---|
+| `OCP_TUI_MAX_CONCURRENT` | `2` | Max concurrent interactive TUI turns. Independent of `CLAUDE_MAX_CONCURRENT` (the stream-json path). Excess turns queue (bounded); a full queue yields a 503. |
+
+---
+
+## Authentication + defunct-reaping (PR-C amendment)
+
+**Date:** 2026-06-13
+**Status:** Accepted — amends ADR 0007.
+**Motivation:** the PI231 production incident — TUI-mode returned `Please run /login · API Error: 401` for days; re-login never stuck.
+
+### How the TUI `claude` authenticates
+
+The spawned interactive `claude` obtains its OAuth bearer in one of two ways, in this order of preference:
+
+1. **`CLAUDE_CODE_OAUTH_TOKEN` in env (PREFERRED).** If the env var is set on the OCP process, `buildTuiCmd` adds `CLAUDE_CODE_OAUTH_TOKEN=<shq-escaped token>` to the pane command's `env` prefix. claude then authenticates via this long-lived token and **never touches the credentials-refresh path**. This is the stable mode — it is exactly how the oracle and Mac-mini hosts already run (and how `server.mjs`'s own `getOAuthCredentials()` takes the same env at highest precedence). cli.js is **not** the authority here: this is a Class B, OCP-owned TUI spawn — see the Class B citation below.
+2. **`<HOME>/.claude/.credentials.json` (FALLBACK).** When the env var is unset, claude falls back to the credentials file and its short-lived access token, renewing via the single-use refresh token.
+
+The token MUST be set explicitly in `buildTuiCmd` because **tmux does not forward the parent process's environment to the pane** (verified live 2026-06-01 — the same reason the whole env is delivered as an `env` prefix). A token sitting in the OCP process env is invisible to the pane unless `buildTuiCmd` re-emits it.
+
+### Why the fallback path corrupts (the PI231 incident)
+
+When the env token is absent, every per-request spawn drives claude through the credentials.json refresh path. OAuth refresh tokens are **single-use / rotating**: a refresh consumes the old refresh token and writes a new one. The per-request `kill-session` teardown can race / interrupt claude mid-rotation, and over many spawn+kill cycles the refresh token ended up an **empty string** — at which point renewal is impossible and the host returns a permanent 401. Re-login writes a fresh token, but the next spawn re-corrupts it. **Proof the env-token fix works:** on the broken PI231 host, `CLAUDE_CODE_OAUTH_TOKEN=<oat01 token> claude -p ...` returned a real answer *despite* the corrupt credentials.json (control without the env token = 401).
+
+**Operator guidance:** set `CLAUDE_CODE_OAUTH_TOKEN` on any TUI-mode host. The credentials.json fallback is retained only for hosts that intentionally rely on it; it is not recommended for a long-running TUI deployment.
+
+**Security note:** with the token in the pane command, it is visible in `ps`. This is acceptable for the **single-user A-path** (it mirrors the existing plaintext-token practice for `server.mjs`), and the **multi-user B-path is already refused at boot** (`CLAUDE_TUI_MODE=true` + `AUTH_MODE=multi` is a hard FATAL), so a guest can never reach this spawn.
+
+### Defunct `<claude>` reaping
+
+The connected leak: the pane's `claude` process is a child of the long-lived **tmux server** daemon, not of the OCP node process (`tmux new-session -d` returns the instant the server forks the pane). Node can therefore never `waitpid()`/reap it — a SIGKILL still needs the *parent* (the tmux server) to reap. `kill-session` destroys the session but leaves the pane's `claude` (and its grandchildren) as `<defunct>` zombies that only the server reaps; over 30 days on PI231 this accumulated to **25 defunct `<claude>`** (a live `tmux kill-server` dropped it 25→3).
+
+The node-reachable action that *actually reaps* — rather than merely re-signalling — is to stop the tmux server: on server exit the kernel reparents survivors to init (PID 1), which reaps them. `reapStaleTuiSessions` therefore, after killing our own `ocp-tui-*` sessions, issues `kill-server` **only when no foreign session of any prefix remains** (coexistence: never disrupt a co-hosted `olp-tui-*` instance). This runs at boot (existing) and now on a 15-min periodic interval gated on TUI-mode and on the TUI path being idle (`inflight === 0 && queued === 0`) so a live turn's pane is never torn down. Residual: a request whose pane is created in the narrow window between the idle-check and `kill-server` would fail cleanly via the existing honesty gates (rare; documented in the server comment).
+
+### ALIGNMENT authorization (Class B)
+
+Both changes are **Class B** (OCP-owned TUI spawn). `cli.js` does not perform either operation — there is no `cli.js` analogue for "how the TUI pane authenticates" or "reaping tmux-server-owned zombies"; this surface is authorized by **this ADR (0007)** per `ALIGNMENT.md`'s Class B citation requirement. No Class A wire surface, no endpoint shape, no `alignment.yml` blacklist token, and no `models.json` entry is touched.
+
+---
+
+## Credential-isolated home for env-token auth (PR-D amendment)
+
+**Date:** 2026-06-13
+**Status:** Accepted — amends ADR 0007. **Corrects** the PR-C rationale and the original "Home strategy" section's scratch-home caveat.
+**Motivation:** PR-C's env-token passing alone did **not** fix the PI231 401. Decisive live evidence (claude 2.1.104, PI231):
+
+| Condition | Result |
+|---|---|
+| env token passed + a broken `~/.claude/.credentials.json` present | **401** (`Please run /login · API Error: 401`) |
+| env token passed + `credentials.json` moved aside | **works** (real answer) |
+
+### Corrected root cause
+
+**Interactive `claude` PREFERS `~/.claude/.credentials.json` over the `CLAUDE_CODE_OAUTH_TOKEN` env var.** A stale/corrupt `credentials.json` therefore **shadows** the env token. (This is *unlike* `-p` mode, where the env token wins — which is why `server.mjs`'s own `getOAuthCredentials()` is unaffected and why PR-C's premise looked sufficient.) So passing the token (PR-C, `buildTuiCmd`) is **necessary but insufficient**: the TUI `claude` must additionally run in a HOME that has **no `credentials.json`**, so the env token is the only credential and is authoritative.
+
+This also fixes the original incident at the **root**, more completely than PR-C claimed: with no `credentials.json` in the home, claude never runs the token-refresh path at all, so the single-use refresh token can never be rotated — and therefore never corrupted — by the spawn+`kill-session` cycle. The 25-zombie / empty-refresh-token failure mode becomes structurally impossible, not merely avoided.
+
+### Decision
+
+When `CLAUDE_CODE_OAUTH_TOKEN` is set, the TUI `claude` runs in a **credential-free scratch home** by default:
+
+- `resolveTuiHome({ realHome, configuredHome, envTokenSet })` (exported from `lib/tui/session.mjs`, pure) decides the home:
+  - **`OCP_TUI_HOME` set** → that path (explicit override, back-compat — an operator who configured it keeps exactly that home).
+  - **else env token set** → `<realHome>/.ocp-tui/home` — a dedicated scratch home seeded with a minimal `.claude.json` (`hasCompletedOnboarding=true` + trust **only** the scratch cwd) and its own `projects/` dir, and **deliberately NO `.credentials.json`** (no symlink, no copy).
+  - **else (no env token)** → the operator's real home — **byte-for-byte the pre-fix behaviour** for hosts that intentionally rely on `credentials.json`.
+- `prepareTuiHome(realHome, tuiHome, cwd, { envTokenMode })` gates the credential handling: in `envTokenMode` it creates the scratch `projects/` dir and seeds the minimal trusted `.claude.json` but **never** creates the credentials symlink. `runTuiTurn` sets `envTokenMode = !!CLAUDE_CODE_OAUTH_TOKEN && ehome !== realHome`.
+- `readTuiTranscript` reads from the **same** home claude runs under (`ehome`), so transcripts land under `<scratch home>/.claude/projects/` and `findTranscriptPath` globs them there — the home is threaded through consistently. (We chose scratch-`HOME` over `CLAUDE_CONFIG_DIR`: the binary supports `CLAUDE_CONFIG_DIR`, but it relocates the transcript root to `<CONFIG_DIR>/projects/` rather than `<HOME>/.claude/projects/`, which would fork the transcript-resolution rule across modes for no benefit. The scratch-HOME lever reuses the existing, tested `prepareTuiHome`/`ehome` plumbing.)
+
+### This RESOLVES — not reintroduces — the scratch-home caveat
+
+The original "Home strategy" section and PR-C's `prepareTuiHome` comment warned that scratch-home is unsafe because *claude rewrites a **symlinked** `.credentials.json` on token refresh → forks/corrupts the OAuth credentials*. **That caveat does not apply to env-token mode**: there is no `credentials.json` in the home to fork, and claude never refreshes (it uses the long-lived env token), so there is no rotation and no corruption. The fork risk was inherent to the *symlink* approach; removing the credentials file entirely removes the risk. The legacy symlink mode is retained **only** for an operator who explicitly sets `OCP_TUI_HOME` without an env token, and its caveat is preserved for exactly that path.
+
+### ALIGNMENT authorization (Class B)
+
+**Class B** (OCP-owned TUI spawn). `cli.js` has no analogue for the TUI pane's auth/home strategy; authorized by **this ADR (0007)** per `ALIGNMENT.md`'s Class B citation requirement. `server.mjs` is touched only to compute `TUI_HOME` via `resolveTuiHome()` (TUI wiring) and to surface the auth mode in the boot log — no Class A wire surface, no endpoint shape, no `alignment.yml` blacklist token, and no `models.json` entry is touched.
+
+---
+
+## Consequences
+
+### Positive
+
+- After 2026-06-15, requests in TUI-mode bill against the Pro/Max subscription pool (`cc_entrypoint=cli`) rather than the Agent SDK credit pool.
+- Kill-switch is immediate (unset env var + restart); zero code change required.
+- Default stream-json path is untouched — no regression risk for existing deployments.
+
+### Negative / trade-offs
+
+- **No token streaming:** responses are buffered then replayed as chunked SSE. Clients see a delay then the full response arrives; real-time token streaming is not available in TUI-mode.
+- **Billing unmeasurable until 2026-06-15:** the `cc_entrypoint=cli` signal is verified, but the credit deduction from the correct pool cannot be confirmed until the billing split activates.
+- **tmux dependency:** the host must have `tmux` installed. CI / Docker images that lack tmux cannot use TUI-mode (the default stream-json path is unaffected).
+- **Wall-clock cap:** long Opus thinking turns may hit the 120 s cap. Increase `CLAUDE_TUI_WALLCLOCK_MS` if needed (no quiescence heuristic — the reader polls until terminal marker or cap).
+- **Grey-area usage:** running an interactive `claude` session headlessly to serve HTTP requests is not an officially documented use case. If Anthropic policy changes to block this pattern, OCP must fall back to the stream-json path (unset `CLAUDE_TUI_MODE`).
+
+### Coexistence
+
+- tmux prefix `ocp-tui-` is registered. Any co-hosted OLP test instance must use `olp-tui-`. Never run two TUI proxies on the same OAuth concurrently — stop one instance during integration testing.
+
+---
+
+## Amendment (2026-07-13) — real SSE streaming via the `MessageDisplay` hook (`OCP_TUI_STREAM`)
+
+**Supersedes**: Request-flow step 6 above ("no real token streaming — deliberate"), for `stream:true`
+requests when `OCP_TUI_STREAM=1`. The buffered path stays the default and is byte-for-byte unchanged.
+
+**Context.** Step 6 was written when the interactive CLI appeared to expose no byte-faithful
+incremental source. A prereq spike (`docs/plans/2026-07-13-tui-latency/streaming-spike.md`) confirmed
+three obvious sources are dead ends — the transcript JSONL grows one *whole event* at a time (the
+answer lands as a single line ~0.3 s before the terminal marker); `tmux capture-pane` yields a
+*rendered* view whose markdown source is unrecoverable (an H2 and a bold span produce identical ANSI);
+`--debug-file` logs stream *timing*, never stream *content*. Every interface that does emit
+`text_delta` (`--output-format stream-json`) requires `-p`, which moves the request to the **metered**
+`sdk-cli` pool — precisely what TUI-mode exists to avoid.
+
+**Decision.** Consume `claude`'s own **`MessageDisplay`** hook, registered via `--settings` on the
+ordinary interactive spawn (no `-p`, no `--bare`). Each fire delivers the **raw markdown source** of an
+incremental `delta` on the hook's stdin. Verified live (claude 2.1.207, sonnet-4-6): banner stays
+`· Claude Max` and the transcript `entrypoint` stays `cli` (subscription pool); `concat(deltas) === T`
+byte-exactly; `T.startsWith(concat(deltas[0..n]))` at every *n*. This is **forwarding, not inventing**
+— ALIGNMENT.md **Class B**. No `cli.js` citation applies: the TUI spawn is OCP-owned surface (this
+ADR), the hook payload is claude's own published contract, and the SSE wire shapes are the OpenAI
+chat/completions streaming spec adopted by **ADR 0006** (the emitters are literally the `-p` path's).
+
+**The transcript remains authoritative.** It is still the terminal-turn signal, still the source of the
+returned/cached text `T`, and still the input to the honesty gates (auth-banner detection C-1,
+`truncated` C-2). The delta stream is a low-latency **mirror**, never a replacement. At end of turn OCP
+asserts the streamed bytes against `T`: equal → serve; a strict *prefix* of `T` → top up from the
+transcript (client still receives exactly `T`); **not** a prefix → **refuse the turn** (SSE error frame,
+no cache, `tui.streamDivergences++`). Serving text the transcript disagrees with is the failure class
+ALIGNMENT.md exists to prevent, so streaming fails loud rather than degrading quietly.
+
+**Consequences / constraints recorded for future authors:**
+
+- **Opt-in, default OFF.** The buffered path is stable production; streaming does not change it.
+- **Per-`session_id` sink is mandatory, not an optimization.** `OCP_TUI_MAX_CONCURRENT` defaults to
+  **2** — two `claude` panes already run concurrently. A single shared sink would interleave one
+  client's deltas into another's stream. The hook writes to `<dir>/<session_id>.jsonl`, the path
+  delivered through the *pane's own env* (`OCP_TUI_STREAM_FILE`); OCP reads only its own turn's file.
+  Verified with two concurrent streamed turns (ALPHA/BRAVO): zero cross-contamination.
+- **Warm-pool compatible (a separate in-flight PR depends on this).** The hook script and the settings
+  file are **static** — nothing request-specific is baked in at spawn time. The sink path derives from
+  the session-id, which for a pre-booted pane is fixed at boot.
+- **The hook is synchronous** (`forceSyncExecution: true` — `claude` *blocks* on it). The hook script
+  must write and exit; it does one `cat` append and nothing else. Measured: p50 **7.2 ms** per fire,
+  ~50 ms across a whole turn — noise against a 6–10 s turn. Do not add work to it.
+- **Thinking blocks do not fire the hook** — verified on a substantive Opus/`xhigh` reasoning turn (see
+  the PR evidence), not merely inferred from the `final:true` call site. This must be **re-verified** if
+  the hook is ever pointed at a new model/effort tier: a thinking delta reaching a client would be
+  unretractable, and the `concat === T` assertion can only *detect* that after the fact, never prevent
+  it. The first-bytes **holdback** (`OCP_TUI_STREAM_HOLDBACK`, default 100 chars) is the same
+  prevention-not-detection reasoning applied to the auth-banner gate.
+- **Block-level granularity**, scaling with answer length — not token-level. Do not promise otherwise.
+- **It moves the first byte, not the last.** Only a progressively-rendering consumer benefits; it does
+  not move TUI-mode's ~6 s TTFT floor.
+
+## Provenance
+
+TUI-mode originated in a prototype contributed via PR #101 (see the PR for author attribution). The productionization design is in `docs/superpowers/specs/2026-05-30-tui-mode-production-design.md`. Spikes S1–S6 / T1–T6 were validated live on the test host against `claude v2.1.158`.

docs/adr/0008-tui-warm-pane-pool.md

@@ -0,0 +1,208 @@
+# ADR 0008 — TUI Warm Pane Pool
+
+**Date:** 2026-07-13
+**Status:** Proposed
+**Extends:** [ADR 0007](0007-tui-interactive-mode.md) (TUI interactive mode). This ADR does not
+change ADR 0007's billing-pool argument, security posture, or kill-switch — it adds a latency
+optimization *inside* the TUI spawn machinery ADR 0007 owns.
+
+---
+
+## Context
+
+TUI mode (ADR 0007) serves every request by cold-booting a fresh `tmux` session running an
+interactive `claude`, submitting one prompt, reading the native transcript, and killing the
+session. That cold boot is paid on **every** request.
+
+[`docs/plans/2026-07-13-tui-latency/`](../plans/2026-07-13-tui-latency/README.md) measured the
+TUI path and listed a warm pane pool as backlog item #3, costed at "**~1.0 s**" (the observed
+boot-to-input-bar time). Instrumenting the real request path showed that estimate is **~4×
+too low**. Phase decomposition of the cold path (n=6 medians, Sonnet 4.6, `--effort low`,
+through a real OCP instance):
+
+| Phase | Median |
+|---|---|
+| prep (trust cwd, write prompt file) | 2 ms |
+| `tmux new-session` | 27 ms |
+| **boot → input bar ready** | **1232 ms** |
+| paste (`load-buffer` + `paste-buffer`) | 8 ms |
+| paste-verify poll | 426 ms |
+| **submit → transcript terminal** | **8458 ms** |
+| teardown | 8 ms |
+| **total** | **10162 ms** |
+| *claude's own reported `turn_duration`* | *5539 ms* |
+| **OCP-side overhead** | **4490 ms** |
+
+The `submit → terminal` phase exceeds claude's own `turn_duration` by **~2.9 s**. That gap is
+**post-input-bar initialization inside `claude`** — work that a pane which has merely *sat idle
+for a few seconds* has already completed. A direct spike confirmed it: an identical pane, idle
+12 s before receiving the same prompt, completed its turn in a median 5537 ms versus 7980 ms
+cold.
+
+So a warm pane recovers **~1.26 s of boot *and* ~2.9 s of in-`claude` cold start** — not the
+~1.0 s the plan predicted.
+
+The reason this was worth a pool rather than a "keep one session and reuse it" cache is a
+hazard already flagged in the code. `lib/tui/transcript.mjs` returns the **last text-bearing
+assistant entry in the whole transcript file**, which is correct *only* under OCP's
+one-session-per-request model, and it says so:
+
+> *"If a future warm-pool ever reuses a session WITHOUT a fresh session-id / clear, earlier-turn
+> text could leak — that author must add user-line scoping here."*
+
+Reusing a pane for a second turn puts two exchanges in one transcript and would leak the earlier
+turn's text into the later turn's answer — a **cross-request data leak**, not merely a bug.
+
+---
+
+## Decision
+
+Add an **opt-in pool of pre-booted, single-use `claude` panes**, `OCP_TUI_POOL_SIZE` (default
+`0` = off, max `4`). Implementation: `lib/tui/pool.mjs`.
+
+### 1. Panes are SINGLE-USE. This is the load-bearing rule.
+
+A pooled pane serves **exactly one turn**, then is killed and replaced in the background. Each
+pane is booted with its **own fresh `--session-id`**, fixed at spawn, and the turn locates its
+transcript by that id.
+
+This preserves one-session-per-request exactly, so the `transcript.mjs` hazard above **does not
+arise** and no user-line scoping was needed. The warning in `transcript.mjs` is deliberately
+left standing, now annotated: it still binds anyone who later wants a pane to serve a second
+turn, or to reset a session with `/clear` and reuse it. **Neither is permitted without first
+adding user-line scoping to the transcript reader.**
+
+Rejected alternative — *reuse a pane for N turns, `/clear` between* — is strictly cheaper
+(no re-boot per request) and was rejected on exactly this basis. The latency win is not worth a
+cross-request text-leak surface guarded only by a `/clear` that we cannot verify landed.
+
+### 2. The pool is keyed by model, and a MISS is always safe.
+
+`--model` is fixed at spawn, so a pane can only serve the model it booted with. A pool miss
+falls back to the existing cold-boot path with **zero behavioural difference**. There is no
+boot-time pre-warm and no configured model: OCP cannot know which model the next caller wants,
+so the pool warms the **most recently requested** model. Consequence, stated plainly: **the
+first request after start, and the first after any model switch, is always a cold miss.**
+
+### 3. The pool and the session reaper coexist by an explicit invariant.
+
+This is the subtle part. `reapStaleTuiSessions()` kills every session matching this instance's
+`ocp-tui-<port>-` prefix, and issues `tmux kill-server` when no foreign session remains (the
+only mechanism that can reap `<defunct>` `claude` zombies — the pane's `claude` is a child of
+the tmux *server*, not of node). A warm pooled pane **is** one of our own sessions, alive and
+idle **by design** — and the periodic sweep runs precisely **when the instance is idle**, i.e.
+exactly when the pool is full.
+
+The invariant, stated in a comment above `reapStaleTuiSessions` and pinned by tests:
+
+1. **A live pooled pane is never reaped — including one that is still BOOTING.** The reaper
+   takes a `spare` set of **exact session names** supplied by the pool's live registry.
+2. **An orphaned pooled pane IS still reaped.** Membership is by **exact name from a live
+   in-memory registry, never by name shape**. A pane the pool no longer owns — handed out,
+   dropped, cancelled, or left behind by a previous process generation (whose registry died with
+   it) — is absent from `spare` and is killed like any other stale session. **Fail-safe:
+   omitting `spare` reaps *more*, never less.** Pool panes are named `ocp-tui-<port>-p<hex>`
+   purely for operator legibility; that shape is *not* the exemption mechanism.
+3. **`kill-server` is suppressed while any pane is spared** (it would kill a live child of the
+   tmux server). Therefore **the pool is DRAINED immediately before every sweep**, so `spare` is
+   empty on the normal tick and `kill-server` still fires. Without the drain, a permanently-full
+   pool would **permanently disable zombie reaping** — the pool would silently break the thing
+   the sweep exists to do. The drain costs one pane re-boot per tick (15 min).
+
+The `spare` mechanism is belt-and-braces given the drain: it makes it impossible for a reap call
+site that *forgets* to drain to kill a live pane.
+
+### 4. The pool tracks its in-flight boot BY NAME, not as a count.
+
+`bootTuiPane` creates the tmux session **synchronously** and only *then* waits (up to
+`POOL_BOOT_MS`, 20 s) for the input bar. So **a pooled tmux session can be live for ~20 s before
+its boot resolves.** A pool that tracked in-flight boots as a *count* could not name that
+session, and this produced two real bugs (both caught in review, both now regression-tested):
+
+- the periodic sweep **killed the booting pane** (it could not be spared), then left the pool
+  empty with nothing scheduled, and logged the exact `tui_pool_boot_failed` warning operators are
+  told to alert on — for a completely healthy drain;
+- graceful shutdown **orphaned a live, authenticated, idle `claude`**: `gracefulShutdown` calls
+  `process.exit(0)` in the same tick as the drain (TUI panes are tmux children, so node's
+  `activeProcesses` set is empty and the "wait for children" path exits immediately), so any
+  cleanup deferred to a `.then()` never ran.
+
+The pool therefore **mints each pane's identity up front** (`{sessionId, name}`, name derived
+from the session-id so `tmux ls` correlates to the transcript file) and holds it in
+`_bootingPane`. `liveNames()` includes it; `drain()` kills it **synchronously**. A generation
+counter distinguishes *"cancelled by us"* from *"genuinely failed"*, so a drain never inflates
+`bootFailures` and `resume()` reliably starts a fresh boot.
+
+### 5. Refills take no concurrency slot, and are serialized.
+
+A refill boot deliberately does **not** take a `TuiSemaphore` slot: those slots bound concurrent
+*turns* and belong to real requests, and charging a background pre-boot against them would let
+the pool starve the traffic it exists to speed up. It cannot leak a slot either, since it never
+holds one. Boots are **serialized** (one at a time): two cold boots racing an in-flight turn were
+observed to overrun even the generous pool readiness cap. A genuinely failed boot does **not**
+re-kick the chain (backoff — a broken `claude` must not respawn forever).
+
+Background boots get a more generous readiness cap (`POOL_BOOT_MS` = 5 × `BOOT_MS`): `BOOT_MS` is
+tight because a *client* is blocked on it, which is not true of a pre-boot. Slow ≠ broken.
+
+---
+
+## Consequences
+
+### Cost — standing processes, paid whether or not a request arrives
+
+**A warm pane is a live idle `claude` process.** Peak process count is
+`OCP_TUI_POOL_SIZE` + `OCP_TUI_MAX_CONCURRENT` + 1 (booting replacement). This is the whole
+reason the pool is **default-off**: an operator must opt into holding processes for traffic that
+may never come. Size is clamped to `POOL_MAX_SIZE` = 4; an unparseable value **disables** the
+pool rather than guessing.
+
+Panes carry a 10-minute TTL and are health-checked at hand-out; a dead or degraded pane becomes
+a **miss** (cold path), never a hung turn.
+
+### Benefit
+
+Measured end-to-end through a real OCP instance (Sonnet 4.6, `--effort low`):
+**p50 10.17 s (n=6, pool off) → 6.00 s (n=12 warm hits) — −4.2 s / −41%.**
+
+### The floor is unchanged
+
+The pool does not touch the **~6 s TTFT floor** documented in the latency plan (claude always
+prefills the full Claude Code system prompt). TUI mode remains unsuitable for interactive /
+real-time consumers; it is for batch and background work. This ADR does not change that
+conclusion.
+
+### Observability
+
+`/health`'s `tui` block gains a `pool` sub-object (`null` when off): `size`, `warm`, `booting`,
+`model`, `hits`, `misses`, `boots`, `bootFailures`, `cancelled`, `dropped`. A climbing
+`bootFailures` means panes are not reaching their input bar — the pool then degrades safely to
+the cold path, but latency reverts to the un-pooled numbers. A steadily climbing `dropped` is
+**normal** (the 15-min sweep drains and re-boots the pool on every tick, by design — see
+Decision 3).
+
+### ALIGNMENT authorization
+
+- **Class B / OCP-owned.** The warm pool is process management around the `claude` CLI — the
+  same category as the existing tmux session lifecycle and the defunct-session reaper it extends.
+  **`cli.js` does not perform this operation, and no `cli.js` citation applies**; the authority
+  is ADR 0007 (which owns the TUI spawn machinery) plus this ADR. This is `ALIGNMENT.md` Rule 2's
+  Class B citation requirement, discharged explicitly rather than by silence.
+- **The `/health` extension** adds sub-fields to the `tui` block. That block is **owned by ADR
+  0007** and post-dates ADR 0006's v3.16.4 grandfather snapshot, so it is not part of the frozen
+  B.2 inventory. The change is additive — every pre-existing `/health` field keeps a
+  byte-identical value, and `pool` is `null` unless the operator opts in — which is the
+  behaviour-preserving bar ADR 0006 sets. This ADR records that authorization.
+- **No spawn argument changed.** `buildTuiCmd` is byte-identical; the pool calls it with the same
+  arguments. Banner-verified on live pooled panes: `· Claude Max`, never `API Usage Billing`
+  (the `--bare` trap documented in the latency plan).
+
+### What a future contributor must not undo
+
+- **Do not let a pane serve a second turn** (or `/clear`-and-reuse one) without first adding
+  user-line scoping to `lib/tui/transcript.mjs`. That is a cross-request text leak, not a perf
+  tweak. See Decision 1.
+- **Do not remove the drain-before-sweep.** It is what keeps `kill-server` zombie reaping alive.
+  See Decision 3.
+- **Do not go back to counting in-flight boots.** The pool must be able to *name* a session that
+  exists but has not finished booting. See Decision 4.

docs/adr/0009-spot-derived-prompt-budget.md

@@ -0,0 +1,54 @@
+# ADR 0009 — Prompt-char budget derives from the models.json SPOT
+
+Date: 2026-07-18
+Status: Accepted (maintainer directive, 2026-07-18: "37.5k 截断未免太短了吧 … 这个在现在还适用吗")
+
+## Context
+
+`MAX_PROMPT_CHARS` (the tail-first truncation guard in `messagesToPrompt`) defaulted to a
+hand-set constant of 150,000 chars ≈ 37.5k English tokens — set in the 200k-window era as a
+runaway-context guard. Meanwhile `models.json` advertises `contextWindow: 200000` for every
+model (and the underlying CLI registry carries 1M native windows for Opus 4.8 / Sonnet 5), and
+`scripts/sync-openclaw.mjs` feeds that 200k into OpenClaw's compaction budget. The result was
+a standing dishonesty identified in the PR #152 review: **no advertised contextWindow value was
+true**, because the proxy silently guillotined every request at ~37.5k tokens — roughly 5×
+below the advertised window — logging only a server-side warning the client never sees.
+
+Raising the constant to another hand-set number would rot the same way. Following the model's
+native 1M directly is also wrong: chars ≠ tokens (CJK runs ~1–1.5 chars/token vs ~4 for
+English, so a 1M-token char cap would let CJK text sail past the model's real window into an
+upstream rejection), single near-window requests can consume a large fraction of a 5-hour
+subscription quota window, and the TUI paste path is untested at megabyte scale.
+
+## Decision
+
+The default budget **derives from the SPOT** instead of being a constant:
+
+```
+MAX_PROMPT_CHARS (default) = max(models.json models[].contextWindow) × 3 chars/token
+                            = 200000 × 3 = 600,000 chars today
+```
+
+Implemented as the pure `derivePromptCharBudget(models, {charsPerToken = 3, floor = 150000})`
+in `lib/prompt.mjs` (unit-tested; floor guards degenerate SPOT states). The multiplier ×3 is
+deliberately conservative: full window for English, and CJK text reaches the model's real
+window at roughly the same point the cap fires — so OCP truncates gracefully (tail-first)
+instead of the upstream rejecting outright.
+
+`CLAUDE_MAX_PROMPT_CHARS` (env) and the runtime settings API remain **absolute overrides**;
+the derivation applies only when neither is set.
+
+## Consequences
+
+- The advertised `contextWindow: 200000` becomes honest: the proxy now actually accepts
+  prompts of that order (English ≈150–200k tokens) before truncating.
+- If `models.json` ever advertises a larger window (e.g. 1M for the 1M-native models), the
+  budget scales automatically — no code change. Whether to advertise 1M is a **separate,
+  deliberate decision** (quota burn per request, OpenClaw compaction memory, TUI paste
+  limits) and is explicitly NOT made by this ADR; the current recommendation is to keep
+  200000 advertised until a real >200k use case appears.
+- One-time behavior change: requests between 150k and 600k chars that were previously
+  truncated now pass through whole — longer TTFT and higher quota consumption for those
+  requests, by design.
+- The truncation mechanism, logging, and the multimodal-path budget threading (PR #154's F2,
+  pending) are unchanged — only the default value's provenance changed.

docs/adr/README.md

@@ -22,6 +22,10 @@ New ADRs increment from the highest existing number. Filenames are
 | [0003](0003-models-json-spot.md) | `models.json` as SPOT | Why model IDs / aliases / context windows live in a single JSON file (not duplicated in `server.mjs` and `setup.mjs` arrays). v3.11.0 refactor. |
 | [0004](0004-openclaw-auto-sync.md) | OpenClaw Auto-Sync | Why `scripts/sync-openclaw.mjs` runs on `ocp update`, what its scope boundary is (writes only `models.providers["claude-local"].models` and `agents.defaults.models["claude-local/*"]`), and the idempotency contract. |
 | [0005](0005-no-multi-provider.md) | No Multi-Provider | Why OCP stays single-provider (Anthropic-via-cli.js) and does not extend to OpenAI / Gemini / OpenRouter. Cost estimate: ~7 weeks for a v1 that buys neither moat nor commercial readiness. Separate commercial work starts in a separate repo. |
+| [0006](0006-openai-shim-scope.md) | OpenAI Shim Scope | The Class A / Class B taxonomy. Class A endpoints (`cli.js`-mirror) keep Rules 1–5 verbatim; Class B endpoints (OCP-owned compatibility surface — `/v1/chat/completions`, `/v1/models`, admin endpoints) are anchored to OpenAI's spec (B.1) or to an authorizing ADR (B.2). Triggered by PR #99 (external `response_format` honoring). Grandfathers the existing B.2 inventory at v3.16.4. |
+| [0007](0007-tui-interactive-mode.md) | TUI Interactive Mode | Why TUI-mode spawns an interactive `claude` in a tmux pane (no `-p`) to reach the **subscription** billing pool (`cc_entrypoint=cli`) rather than the metered Agent SDK pool. Owns the TUI spawn machinery: entrypoint labeling, credential-isolated home, MCP hard-disable, session namespace + defunct-session reaping, the independent concurrency bound, and the `/health` `tui` block. **Single-user only** — hard FATAL on multi-user configs. |
+| [0008](0008-tui-warm-pane-pool.md) | TUI Warm Pane Pool | Why `OCP_TUI_POOL_SIZE` pre-boots **single-use** `claude` panes (one turn each, own `--session-id`) — and why reuse is forbidden (`transcript.mjs` returns the last assistant entry in the file, so a reused session leaks the earlier turn's text). Measured −41% end-to-end. Defines the pool↔reaper invariant (exemption by exact name from a live registry; drain before every sweep so `kill-server` zombie reaping survives) and the standing idle-process cost. Extends ADR 0007. |
+| [0009](0009-spot-derived-prompt-budget.md) | SPOT-Derived Prompt Budget | Why `MAX_PROMPT_CHARS`'s default is `max(models.json contextWindow) × 3 chars/token` (600k chars today) instead of a hand-set constant — the old 150k silently under-delivered the advertised window ~5×. ×3 is the CJK-safe multiplier; env/settings stay absolute overrides; whether to advertise 1M windows is explicitly a separate decision. |
 
 ## When to write a new ADR
 

docs/lan-mode.md

@@ -0,0 +1,396 @@
+Part of [OCP](../README.md) — LAN & multi-user: server setup, client connect, API-key management, per-key quotas, anonymous access, and the deployment/security model (including the honest limits of sharing).
+
+# LAN & multi-user
+
+OCP has two roles: **Server** (runs the proxy, needs Claude CLI) and **Client** (connects to a server, zero dependencies).
+
+```
+┌─ Server (always-on device) ─────────────────────────────┐
+│  Mac mini / NAS / Raspberry Pi / Desktop                │
+│  Claude CLI + OCP server → bound to 0.0.0.0:3456       │
+└───────────────────────┬─────────────────────────────────┘
+                        │ LAN
+    ┌───────────────────┼───────────────────┐
+    ▼                   ▼                   ▼
+ Laptop             Phone/Tablet        Pi / Server
+ (client)           (browser)           (client)
+```
+
+## Server Setup
+
+> **Recommended:** Install OCP on a device that stays powered on — Mac mini, NAS, Raspberry Pi, or a desktop that doesn't sleep. This ensures all clients always have access.
+
+**Prerequisites:**
+- macOS or Linux (Windows is not supported — `setup.mjs` installs launchd / systemd auto-start)
+- Node.js 22.5+ (Node 23+ recommended — `node:sqlite` is fully stable without flags from 23.0; on 22.5–22.x it works behind `--experimental-sqlite`)
+- `git`
+- [Claude CLI](https://docs.anthropic.com/en/docs/claude-cli) — install and authenticate:
+  ```bash
+  npm install -g @anthropic-ai/claude-code
+  claude auth login   # prints a URL + code — open URL on any browser, sign in, paste code back
+  ```
+  Headless servers (Pi / NAS / VPS without a desktop browser): see [Headless install notes](#headless-install-notes) below.
+
+```bash
+# 1. Clone and run setup
+git clone https://github.com/dtzp555-max/ocp.git
+cd ocp
+node setup.mjs
+```
+
+The setup script will:
+1. Verify Claude CLI is installed and authenticated
+2. Start the proxy on port 3456
+3. Install auto-start (launchd on macOS, systemd on Linux)
+
+After install the `ocp` CLI lives at `~/ocp/ocp`. To put it on your PATH, either symlink it manually (`ln -sf ~/ocp/ocp ~/.local/bin/ocp` if `~/.local/bin` is on your PATH, or `sudo ln -sf ~/ocp/ocp /usr/local/bin/ocp` for a system-wide symlink) or add an alias (`alias ocp=~/ocp/ocp`). Otherwise invoke it as `~/ocp/ocp <subcommand>`. The rest of this document assumes `ocp` is on your PATH.
+
+> **Cloud/Linux servers:** If `ocp: command not found` after a cloud install, the binary isn't in PATH. Full path in that layout: `~/.openclaw/projects/ocp/ocp`
+
+**Single-machine use** — just set your IDE to use the proxy:
+```bash
+export OPENAI_BASE_URL=http://127.0.0.1:3456/v1
+```
+
+**LAN mode** — reach OCP from your own devices on the network (Claude Pro/Max are per-user accounts — see [Sharing with family / a team — honest limits](#deployment-model--security-read-this) before extending access to other people):
+```bash
+# Enable LAN access with per-user auth (recommended)
+node setup.mjs --bind 0.0.0.0 --auth-mode multi
+```
+
+Then create API keys for each person/device:
+```bash
+# Generate a strong admin key (one-time — save it for later key management):
+export OCP_ADMIN_KEY=$(openssl rand -base64 32)
+# Add the same export line to ~/.zshrc or ~/.bashrc so it persists.
+
+ocp keys add wife-laptop
+#  ✓ Key created for "wife-laptop"
+#    API Key: ocp_example12345abcde...
+#    Copy this key now — you won't see it again.
+
+ocp keys add son-ipad
+ocp keys add pi-server
+```
+
+Run `ocp lan` to see your IP and ready-to-share instructions.
+
+**Verify:**
+```bash
+curl http://127.0.0.1:3456/v1/models
+# Returns: claude-opus-4-8, claude-opus-4-7, claude-opus-4-6, claude-sonnet-5, claude-sonnet-4-6, claude-haiku-4-5-20251001
+```
+
+### Headless install notes
+
+OCP is designed for always-on devices that often don't have a desktop browser — Mac mini, NAS, Raspberry Pi, cloud VPS. The Claude CLI auth flow still works headless:
+
+**Option 1 — interactive OAuth over SSH (one-shot).** `claude auth login` prints a URL + 8-digit code. Open the URL on **any** device with a browser (your laptop, phone), sign in to your Anthropic account, and paste the code back into the SSH session. No browser needed on the server itself.
+
+**Option 2 — long-lived token (auth once, no re-prompts).**
+
+```bash
+claude setup-token   # subscription-backed long-lived token
+```
+
+Same Claude subscription as Option 1; the token is stored in Claude CLI's normal config location. Useful when you'd rather not redo the OAuth flow when sessions expire.
+
+If `claude auth login` errors out with something like `cannot open browser`, you've hit the same case — fall back to either option above.
+
+## AI-assisted install prompts
+
+If you've got Claude Code, Cursor, or any other AI coding assistant on this machine, you can copy-paste one of these prompts and let the AI walk through the install for you. Each prompt pins the AI to the right README section, names the verification step, and forbids silent retries — so you stay in the loop.
+
+**Single-machine use** — install OCP for IDEs on this same machine only:
+
+```text
+I want to install OCP on this machine to use my Claude Pro/Max subscription
+as an OpenAI-compatible API for local IDEs.
+
+Please follow https://github.com/dtzp555-max/ocp/blob/main/README.md
+§Quickstart (single-machine install):
+
+1. Verify prerequisites: macOS or Linux, Node.js 22.5+, git, Claude CLI
+   installed and logged in (`claude auth status`). Install missing pieces
+   using my system's package manager.
+2. git clone the repo, cd in, and run `node setup.mjs`.
+3. Verify with `curl http://127.0.0.1:3456/v1/models` (should list 6 models).
+4. Add `export OPENAI_BASE_URL=http://127.0.0.1:3456/v1` to my shell rc.
+5. Tell me to reload my shell and try a tool like Cline / Continue / Cursor.
+
+Before each step, tell me what you'll run and wait for confirmation.
+On any error, diagnose first — don't auto-retry.
+```
+
+**LAN mode (server)** — install OCP as a server so your own devices on the LAN can reach it (Claude Pro/Max are per-user accounts — review Anthropic's Usage Policy before extending access to other people):
+
+```text
+I want to install OCP on this device as a LAN server so my own devices on the
+network can reach my Claude Pro/Max subscription through a local
+OpenAI-compatible endpoint.
+
+Please follow https://github.com/dtzp555-max/ocp/blob/main/docs/lan-mode.md
+"Server Setup" → "LAN mode" path:
+
+1. Verify prerequisites: macOS or Linux (Windows not supported), Node.js
+   22.5+, git, Claude CLI installed and authenticated.
+2. Generate a strong admin key with `openssl rand -base64 32`. Save it —
+   I'll need it to manage per-user keys later.
+3. git clone https://github.com/dtzp555-max/ocp.git && cd ocp
+4. Run `node setup.mjs --bind 0.0.0.0 --auth-mode multi`.
+5. Add OCP_ADMIN_KEY to my shell rc (~/.zshrc or ~/.bashrc).
+6. Run `ocp lan` to show me the LAN IP and connect command.
+7. Optionally create example keys: `ocp keys add laptop`, `ocp keys add tablet`.
+8. Verify: `curl http://127.0.0.1:3456/v1/models` returns 6 models.
+
+Tell me each step before running it. On error, diagnose before retrying.
+```
+
+**Client connect** — configure this device to use an existing OCP server on your LAN:
+
+```text
+There's an OCP server at <SERVER_IP> on my LAN. Configure this machine to
+use it for any local IDEs (Cursor, Cline, Continue.dev, OpenCode, OpenClaw).
+
+Server IP: <SERVER_IP>
+API key (leave blank if the server has anonymous mode enabled): <OPTIONAL_KEY>
+
+Please follow https://github.com/dtzp555-max/ocp/blob/main/docs/lan-mode.md
+"Client Setup" path:
+
+1. Download ocp-connect:
+     curl -fsSL https://raw.githubusercontent.com/dtzp555-max/ocp/main/ocp-connect -o ocp-connect
+     chmod +x ocp-connect
+2. Run `./ocp-connect <SERVER_IP>` (add `--key <KEY>` if you have one).
+3. Follow any IDE-specific manual hints it prints.
+4. Verify: `curl http://<SERVER_IP>:3456/v1/models` returns 6 models.
+5. Tell me to reload my shell + restart any IDE that was already running.
+
+Don't auto-retry on error. Tell me the failure mode first.
+```
+
+## Client Setup
+
+> Clients do **not** need to install Node.js, Claude CLI, or the OCP repo. Only `curl` and `python3` are required (pre-installed on most Linux/Mac systems).
+>
+> **Find the server's LAN IP** by running `ocp lan` on the server machine — it prints both the IP and a ready-to-share connect command.
+
+**One-command setup** — download the lightweight `ocp-connect` script:
+
+```bash
+curl -fsSL https://raw.githubusercontent.com/dtzp555-max/ocp/main/ocp-connect -o ocp-connect
+chmod +x ocp-connect
+./ocp-connect <server-ip>
+```
+
+**Zero-config** — when the server admin has set `PROXY_ANONYMOUS_KEY` *and* opted in with `PROXY_ADVERTISE_ANON_KEY=1` (see [Anonymous Access](#anonymous-access-optional) below), just pass the server IP and nothing else. `ocp-connect` reads the anonymous key from `/health` and uses it automatically. Without the opt-in, `/health` does not expose the key (issue #109); pass `--key` or rely on anonymous access instead:
+
+```bash
+./ocp-connect <server-ip>
+```
+
+If the server requires a key, pass it with `--key`:
+```bash
+./ocp-connect <server-ip> --key <your-api-key>
+```
+
+Or as a one-liner (no file saved):
+```bash
+curl -fsSL https://raw.githubusercontent.com/dtzp555-max/ocp/main/ocp-connect | bash -s -- <server-ip>
+```
+
+Example:
+```
+$ ./ocp-connect 192.168.1.100
+
+OCP Connect v1.3.0
+─────────────────────────────────────
+  Remote: http://192.168.1.100:3456
+
+  Checking connectivity...
+  ✓ Connected
+
+  Remote OCP v3.11.0  (auth: multi)
+
+  ⓘ Using server-advertised anonymous key: ocp_publ...n_v1
+    (set by admin via PROXY_ANONYMOUS_KEY; see issue #12 §14 Path A)
+
+  Testing API access...
+  ✓ API accessible (6 models available)
+
+  Shell config:
+    ✓ .bashrc
+    ✓ .zshrc
+    OPENAI_BASE_URL=http://192.168.1.100:3456/v1
+
+  System-level (launchctl):
+    ✓ OPENAI_BASE_URL set for GUI apps and daemons
+
+  IDE Configuration
+  ─────────────────────────────────────
+  Detected: OpenClaw (~/.openclaw/openclaw.json)
+
+  Configure OpenClaw to use this OCP? [Y/n] y
+  Provider name (models show as <name>/model-id) [ocp]: ocp
+
+  How should OCP models be configured?
+    1) Primary — use OCP by default, keep existing models as backup
+    2) Backup  — keep current primary, add OCP as additional option
+
+  Choice [1]: 1
+
+  Writing OpenClaw config...
+    ✓ Per-agent auth profile seeded (2):
+      • ~/.openclaw/agents/main/agent/auth-profiles.json
+      • ~/.openclaw/agents/macbook_bot/agent/auth-profiles.json
+  ✓ OpenClaw configured
+    Provider: ocp
+    Models:
+      • ocp/claude-opus-4-8
+      • ocp/claude-opus-4-7
+      • ocp/claude-opus-4-6
+      • ocp/claude-sonnet-5
+      • ocp/claude-sonnet-4-6
+      • ocp/claude-haiku-4-5-20251001
+    Priority: PRIMARY (default model)
+
+    Restart OpenClaw to apply: openclaw gateway restart
+
+  Running smoke test...
+  ✓ Smoke test passed: OK
+    Note: smoke test only verifies OCP is reachable and the key is valid.
+    It does not verify your IDE/agent end-to-end. To verify OpenClaw works,
+    restart it (`openclaw gateway restart`) and send a test message to your bot.
+
+  Done. Reload your shell to apply:
+    source ~/.zshrc
+```
+
+The script automatically:
+- Writes env vars to all relevant shell rc files (`.bashrc`, `.zshrc`)
+- Sets system-level env vars (`launchctl setenv` on macOS, `environment.d` on Linux)
+- **Auto-discovers anonymous key** from `/health.anonymousKey` when no `--key` given (v1.3.0+, requires server v3.10.0+; server must also set `PROXY_ADVERTISE_ANON_KEY=1` — see [Anonymous Access](#anonymous-access-optional))
+- Configures OpenClaw automatically (including per-agent `auth-profiles.json` for multi-agent setups)
+- Detects Cline, Continue.dev, Cursor, and opencode, and prints setup hints (manual configuration required for these IDEs)
+
+On macOS, `launchctl setenv` vars reset on reboot — re-run `ocp-connect` after restart.
+
+**Manual setup** — if you prefer not to use the script:
+```bash
+export OPENAI_BASE_URL=http://<server-ip>:3456/v1
+export OPENAI_API_KEY=ocp_<your-key>
+```
+Add these lines to `~/.bashrc` or `~/.zshrc` to persist across sessions.
+
+## Monitoring (Server-side)
+
+```bash
+# Per-key usage stats
+ocp usage --by-key
+#  Key                  Reqs   OK  Err  Avg Time
+#  wife-laptop             5    5    0      8.0s
+#  son-ipad                3    3    0      6.2s
+
+# Manage keys
+ocp keys              # List all keys
+ocp keys revoke son-ipad   # Revoke a key
+```
+
+**Web Dashboard:** Open `http://<server-ip>:3456/dashboard` in any browser for real-time monitoring — per-key usage, request history, plan utilization, and system health.
+
+![OCP Dashboard](images/dashboard.png)
+
+## Auth Modes
+
+| Mode | Env | Use Case |
+|------|-----|----------|
+| `none` | `CLAUDE_AUTH_MODE=none` | Trusted home network, no auth needed |
+| `shared` | `CLAUDE_AUTH_MODE=shared` + `PROXY_API_KEY=xxx` | Everyone shares one key |
+| `multi` | `CLAUDE_AUTH_MODE=multi` + `OCP_ADMIN_KEY=xxx` | Per-person keys for usage tracking + quotas (trusted users only — see Deployment model below) |
+
+> **Usage scope (v3.14.0+):** `/api/usage` returns the caller's own rows by default. Admin callers must pass `?all=true` to retrieve data for all keys; doing so emits an audit log line.
+
+## Deployment model & security (read this)
+
+**What OCP is built for today: single-user, multi-IDE.** Run OCP as a server on one machine and point all of *your own* IDEs/devices at it — one Claude Pro/Max subscription, used everywhere. This is the primary, solid use case.
+
+**Sharing with family / a team — honest limits.** You *can* share OCP on a LAN, but be clear about what the auth modes do and don't give you:
+
+- The per-key modes (`shared` / `multi`) give per-key **usage tracking, quotas, and cache separation** — useful for seeing who used what and capping budgets.
+- They do **not** give a **security isolation boundary**. The spawned `claude` runs with the **operator's filesystem access** and is *not* sandboxed per key. **Only share with people you fully trust, on a trusted network.**
+- For simple trusted family sharing, the easiest setup is a single shared **anonymous key** (see [Anonymous Access](#anonymous-access-optional)) — no per-person separation, same trust assumption.
+- **Account terms and ToS — read before sharing with others.** Claude Pro/Max are *per-user* accounts. Pooling a single subscription across **multiple distinct people** may violate Anthropic's Consumer Terms of Service and risk account suspension by the abuse classifier. The defensible framing is **"one person, your own devices"** — sharing with friends or a team is not. OCP does not change your account terms, and whether any particular sharing setup complies with the ToS is the account holder's responsibility. Review Anthropic's Usage Policy before extending access to other people.
+
+**Real per-user isolation (sandboxed, multi-tenant-safe) is planned for after 2026-06-15** — per-key ephemeral home + tool lockdown + an OS sandbox. Until then, treat a multi-user OCP as a *trusted-group convenience*, not a security boundary. (This is also why `CLAUDE_TUI_MODE` is single-user-only — see [Subscription-pool (TUI) mode](tui-mode.md#subscription-pool-tui-mode).)
+
+## Anonymous Access (optional)
+
+In `multi` mode, the admin can designate a single well-known "anonymous" key that bypasses `validateKey()` and grants public read/write access. This is useful for letting LAN users (or clients like OpenClaw multi-agent setups) connect without individual per-user keys.
+
+**Enable**:
+
+The anonymous key is wired into the service unit (launchd plist on macOS, systemd unit on Linux) at install time. Export `PROXY_ANONYMOUS_KEY` in your shell before running `setup.mjs`, and `setup.mjs` will write it into the service unit env so the auto-started proxy picks it up:
+
+```bash
+export PROXY_ANONYMOUS_KEY=ocp_public_anon   # or any string of your choice
+node setup.mjs --bind 0.0.0.0 --auth-mode multi
+```
+
+If OCP is already installed without it, re-export the env var and re-run `node setup.mjs` (the installer is idempotent — it refreshes the service unit). Then `ocp restart` so the running proxy picks up the new env. Setting `PROXY_ANONYMOUS_KEY` only in your interactive shell **does not** affect the auto-started proxy — the service unit is the source of truth for its environment.
+
+**Client side**: the anonymous key value is exposed via `GET /health` as the field `anonymousKey` (null when not set) **only to localhost callers** or when the admin has also set `PROXY_ADVERTISE_ANON_KEY=1` (default off — see issue #109). With that opt-in, clients like `ocp-connect` can auto-discover and use it, so the end user doesn't need to get a personal key from the admin.
+
+**Security note**: setting this env var is an **opt-in** to public access — anyone who can reach your OCP endpoint can use it, up to any rate limits you configure. Don't enable this on internet-exposed OCP instances without additional protection.
+
+**Not a secret**: because `/health` is an unauthenticated endpoint, the anonymous key is **publicly readable** by anyone who can reach the server. That is intentional — the key exists so clients can self-configure without out-of-band coordination. Treat it as a convenience handle, not as an access credential.
+
+## Per-Key Quota (Budget Control)
+
+Prevent any single user from exhausting your subscription. Set daily, weekly, or monthly request limits per API key:
+
+```bash
+# Set a daily limit of 50 requests for a key
+curl -X PATCH http://127.0.0.1:3456/api/keys/wife-laptop/quota \
+  -H "Authorization: Bearer $OCP_ADMIN_KEY" \
+  -d '{"daily": 50}'
+
+# Set multiple limits at once
+curl -X PATCH http://127.0.0.1:3456/api/keys/son-ipad/quota \
+  -H "Authorization: Bearer $OCP_ADMIN_KEY" \
+  -d '{"daily": 20, "weekly": 100}'
+
+# Check current quota + usage
+curl http://127.0.0.1:3456/api/keys/wife-laptop/quota
+# → { "daily": { "limit": 50, "used": 12 }, "weekly": { "limit": null, "used": 34 }, ... }
+
+# Remove a limit (set to null)
+curl -X PATCH http://127.0.0.1:3456/api/keys/wife-laptop/quota \
+  -d '{"daily": null}'
+```
+
+When a key exceeds its quota, OCP returns HTTP 429 with a structured error:
+```json
+{
+  "error": {
+    "message": "Quota exceeded: 50/50 requests (daily). Resets 6h 12m.",
+    "type": "quota_exceeded",
+    "quota": { "period": "daily", "limit": 50, "used": 50, "resetsIn": "6h 12m" }
+  }
+}
+```
+
+- `null` = unlimited (default for all keys)
+- Only successful requests count toward quota
+- Admin and anonymous users are never subject to quotas
+- PATCH is a partial update — omitted fields are left unchanged
+
+> **Note:** quotas are best-effort. Under concurrent bursts a key can exceed its cap by up to the server's max-concurrency (default 8), and cache hits are not counted toward quota. They cap budgets for cooperative family use, not adversarial abuse.
+
+## Important Notes
+
+- All users share your Claude Pro/Max **rate limits** (5h session + 7d weekly)
+- `ocp usage` shows how much quota remains
+- Keys are stored in `~/.ocp/ocp.db` (SQLite, zero external dependencies)
+- Admin key is required for key management API endpoints
+- The dashboard (`/dashboard`) and health check (`/health`) are always public
+- File modes for `~/.ocp` (0700), `admin-key` + `ocp.db` (0600) are auto-tightened at server startup as of v3.14.0

docs/openai-compat-pin.md

@@ -0,0 +1,11 @@
+# OpenAI Compatibility Pin (Class B.1)
+
+**Status:** Placeholder — populated at first B.1 audit per ADR 0006 §"Class B audit cadence".
+
+This file is the Class B.1 counterpart to the Class A `cli.js` audit pin in `ALIGNMENT.md` §"Golden Reference". When the first annual alignment audit covers Class B (per `ALIGNMENT.md` §"Annual Alignment Audit"), this file will be populated with:
+
+- The OpenAI `/v1/chat/completions` specification snapshot date being audited against (and the source URL the snapshot was taken from).
+- The list of B.1 endpoints (currently `/v1/chat/completions`, `/v1/models`) and, for each one, the specific OpenAI spec fields and behaviours it honors.
+- Drift detection notes for any OpenAI spec changes since the previous audit, and any OCP code changes required to track those changes.
+
+Until populated, this file's existence is only a forward reference so that the link in `ALIGNMENT.md` does not 404. The actual audit procedure is defined in `ALIGNMENT.md` §"Annual Alignment Audit" (Class B scope) and ADR 0006 §"Class B audit cadence".

docs/plans/2026-07-13-tui-latency/README.md

@@ -0,0 +1,236 @@
+# TUI-mode latency: measured floor, and the four things worth fixing
+
+**Date**: 2026-07-13
+**Status**: findings + backlog. **Superseded in part** — see the dated update boxes below.
+Item #1 shipped ([#156](https://github.com/dtzp555-max/ocp/pull/156)); item #2 is **dead**
+([`streaming-spike.md`](streaming-spike.md)); item #4 measured, **no effect**; item #3 stands.
+**Measured on**: Mac mini / macOS 26.5.2 / Claude Code **v2.1.207** / Sonnet 5 / Claude Max subscription / **real-home mode** (no `CLAUDE_CODE_OAUTH_TOKEN`, no `OCP_TUI_HOME` in the service env)
+**Evidence**: [`measurements.jsonl`](measurements.jsonl) — **n=15** (3 configs × 5) · banner captures [`billing-banner.txt`](billing-banner.txt) · harness [`floor.sh`](floor.sh)
+
+## Why this exists
+
+An external consumer (the 知音 AI project) benchmarked OCP's prompt path and measured
+**TTFT p50 ≈ 30–32 s**, and excluded OCP as a backend on that basis. That number is real,
+but it is *not* the model being slow — this document decomposes where the 30 seconds
+actually go, and what OCP can do about it.
+
+**The harness deliberately does not go through OCP.** It spawns `tmux` + `claude` directly
+(session prefix `zhiyin-floor-`, never `ocp-tui-*`) and polls `tmux capture-pane` for
+incremental render, so it measures the **true first-token time** of the underlying
+subscription path — the floor OCP could reach if it were perfect.
+
+---
+
+## Measurements
+
+All rows in [`measurements.jsonl`](measurements.jsonl); every number below is recomputable from it.
+
+| Config | n | boot→input-ready (median) | **TTFT (median)** | TTFT range | full answer (median) |
+|---|---|---|---|---|---|
+| baseline (inherits global `effortLevel: xhigh`) | 5 | 1.07 s | **10.35 s** | 8.32 – 17.19 s | 11.32 s |
+| **`--effort low`** | 5 | 1.03 s | **6.17 s** | **5.87 – 6.44 s** | 9.98 s |
+| `--bare` | 5 | 0.44 s | **no answer at all** (5/5 `ttft_ms: -1`) | — | — |
+
+> **Not from this harness**: the direct Anthropic API reference figure (TTFT 0.84–1.64 s, n=2)
+> comes from the 知音 AI project's own smoke test, not from `measurements.jsonl`. It is quoted
+> only to size the gap; do not look for it in the evidence file.
+
+### Where the 30 seconds go
+
+```
+ ~1.0 s   spawn → claude's input bar is ready          ← NOT the bottleneck
+ ~6-10 s  true TTFT (first token rendered in the pane)
+ ~20 s    ████ waiting for the whole turn to finish ████  ← this is the 30s
+```
+
+`runTuiTurn` blocks on the native transcript until a terminal event (`lib/tui/session.mjs`
+"Block on the native transcript … until terminal"; `readTuiTranscript` in
+`lib/tui/transcript.mjs`; ADR 0007 step 4) — i.e. it waits for the **entire turn** to complete
+before returning anything. There is no streaming path. The ~20 s delta between this harness's
+real TTFT and OCP's reported 30–32 s is exactly that.
+
+> **⚠️ 2026-07-13 correction — this decomposition attributes the ~20 s to the wrong thing.** It was
+> inferred from the external 30–32 s report, never measured *through* OCP. It has since been measured
+> through a real OCP instance (TUI mode, `claude-sonnet-4-6`, the same ~1850-token prompt, n=5):
+> **median 11.30 s** before [#156](https://github.com/dtzp555-max/ocp/pull/156), **9.55 s** after.
+> Same-turn decomposition (baseline row `i=5`): **11.563 s** wall through OCP vs `turn_duration:
+> 7.319 s` of CLI-internal time on that same turn → **OCP's own overhead ≈ 4.2 s** (n=1), **not
+> ~20 s**. The rest of any larger number is the model *generating a long answer*,
+> which the blocking wait does not cause and streaming would not shorten — it would only move the
+> first byte earlier. The 30–32 s figure therefore reflects a much longer output (and/or the
+> then-inherited `xhigh` effort), not 20 s of OCP dead time. See
+> [`streaming-spike.md`](streaming-spike.md) § "What streaming would have bought".
+
+---
+
+## ⚠️ Blocking constraint: `--bare` silently drops you off the subscription pool
+
+Captured live ([`billing-banner.txt`](billing-banner.txt)) — the startup banner is the **only**
+reliable indicator:
+
+```
+[]                     | Sonnet 5 with xhigh effort · Claude Max
+[--effort low]         | Sonnet 5 with low effort  · Claude Max
+[--bare]               | Sonnet 5 with xhigh effort · API Usage Billing     ← ❌
+```
+
+`--bare` ("skip hooks, LSP, plugin…") **also skips the subscription-credential resolution
+path**. It really does cut boot to 0.43–0.45 s — but you are no longer on the subscription,
+which defeats the entire purpose of TUI mode (ADR 0007 exists solely to reach the
+subscription pool).
+
+**The failure is silent.** All 5 `--bare` samples reached input-ready (boot 0.43–0.45 s), were
+sent the prompt, and then produced **no answer at all** — 60 s timeout, no error, no crash, the
+pane simply never rendered a token (the API-billing account had no credit balance). Nothing in
+the transcript or the exit status reveals this.
+
+**Anyone changing spawn flags must diff the banner line before and after.**
+
+---
+
+## Backlog — four items, ranked by value ÷ effort
+
+### 1. Pass `--effort` explicitly on spawn — **do this first**
+
+`buildTuiCmd` (`lib/tui/session.mjs`) does not pass `--effort` — `grep -rn -- "--effort\|effortLevel" lib/ server.mjs`
+returns zero hits. What the pane's `claude` ends up using therefore depends on **which HOME mode
+`resolveTuiHome()` picked**:
+
+| mode | HOME | effort the pane gets |
+|---|---|---|
+| **real-home** (legacy default — *current* service config: no `CLAUDE_CODE_OAUTH_TOKEN`, no `OCP_TUI_HOME`) | `~` | **inherits the operator's `~/.claude/settings.json` → `effortLevel: xhigh` on this host** |
+| env-token scratch (`CLAUDE_CODE_OAUTH_TOKEN` set — the direction #146/#150 pushed) | `~/.ocp-tui/home` | that settings.json contains only `permissions.additionalDirectories`; `prepareTuiHome()` never writes `effortLevel` → **claude's built-in default** |
+
+**Scope note**: TUI mode is currently *off* on this host (`CLAUDE_TUI_MODE=false`; `/health` →
+`"tui": {"enabled": false}`), so live traffic takes the `-p` path today. The statement below is
+about what happens **when TUI mode is enabled**.
+
+On the current HOME config, **every TUI request would run extended thinking** — pure waste
+for the typical "generate this JSON" request, and it makes latency depend on an unrelated global
+setting the operator may have changed for their own interactive use. And the mode split means
+the effort level silently changes if the operator ever switches to env-token mode.
+**Passing `--effort` explicitly fixes both problems at once.**
+
+- **Effect (real-home, measured)**: TTFT p50 **10.35 s → 6.17 s (−40 %)**, and the spread
+  collapses from 8.32–17.19 s to **5.87–6.44 s**. For a proxy, the variance reduction matters
+  more than the median.
+- **Cost**: one flag. Suggested: a new `OCP_TUI_EFFORT` env var (default `low`), documented in
+  README § "Environment Variables" per `release_kit.new_feature_doc_expectations`.
+- **Risk**: none — banner confirms it stays on `Claude Max` (see `billing-banner.txt`).
+- ⚠️ Do **not** reach for `--bare` to shave boot: see above.
+
+### 2. Real streaming instead of blocking on turn-terminal — **ACHIEVABLE → [`streaming-spike.md`](streaming-spike.md)**
+
+> **2026-07-13 update — the prereq spike was run. The answer is YES, but not from either source this
+> item guessed at.** (a) The transcript grows at *event* granularity (the whole answer lands in one
+> line, ~0.3 s before terminal) — dead. (b) The pane is a **rendered** view whose `capture-pane` text
+> no longer contains the answer's source bytes (`## `, `**`, code fences are gone) — dead, and worse
+> than "lossy": it is *not the model's text*. **But there is a third source neither this backlog nor
+> the first spike considered: `claude` fires a `MessageDisplay` hook carrying incremental,
+> byte-faithful `delta`s of the raw reply.** Verified live on a plain interactive TUI spawn (no `-p`),
+> banner `· Claude Max`: 7 fires spread across generation, `concat(deltas) === T` **byte-exactly**
+> (579 == 579), `T.startsWith(S)` true at every step, `## ` / `**` / ```` ```javascript ```` all
+> present in the deltas. Granularity is block-level (~5–7 chunks/answer), not token-level — plenty for
+> SSE. **Build it.**
+>
+> ⚠️ Two corrections to this item as written: the **"~20 s" is wrong** (inferred from an external
+> report, never measured through OCP — the same-turn decomposition puts OCP's own overhead at **~4 s**,
+> n=1), and **streaming moves the first byte, not the last** — so a consumer needing the *complete*
+> answer (the JSON-card case that motivated this) gains **nothing** from it. Build it for
+> progressively-rendering consumers, not as a throughput win.
+>
+> Full evidence + implementer caveats (the hook is `forceSyncExecution` — claude BLOCKS on it):
+> **[`streaming-spike.md`](streaming-spike.md)**. Original framing preserved below.
+
+Today `runTuiTurn` blocks on the transcript until the turn is *finished*. The pane is already
+rendering tokens incrementally the whole time — this harness proves you can observe first token
+at ~6 s by polling `tmux capture-pane`.
+
+- **Effect**: turns a 30 s wall into a ~6 s TTFT with progressive output; enables SSE streaming
+  on the OCP endpoint instead of a single blob at the end.
+- **Cost**: real work. Pane capture is ANSI/redraw-based and lossy for exact text (wrapping,
+  scrollback, spinner lines). Two candidate sources: (a) incremental reads of the transcript
+  JSONL, (b) `capture-pane` diffing with a stable start marker. (a) is much cleaner **if it
+  holds**.
+- **Prereq spike (do this before designing anything)**: does the transcript JSONL grow *during*
+  a turn, or only at the end? If only at the end, (a) is dead and you are stuck with (b).
+
+### 3. Warm pane pool — ~1 s
+
+Every request spawns a fresh tmux session + `claude` (`randomUUID()` + `new-session`, then
+`kill-session` in `finally`; `grep -rn "pool\|warm\|reuse" lib/tui/*.mjs` → zero hits). Boot to
+input-ready is ~1.0 s, paid on every request. A pool of pre-booted panes (single-use, replaced in
+the background) amortizes it to zero for any workload below the pool refill rate.
+
+- **Effect**: −1.0 s.
+- **Cost**: moderate; interacts with the session reaper and the per-port prefix scoping added in
+  #148 — pooled panes must not look like zombies to the sweep.
+- Lower priority than #1 and #2: it is the smallest slice.
+
+### 4. Trim the prefill — ~~probably not worth it~~ **MEASURED: no detectable benefit. Do not adopt.**
+
+> **2026-07-13 update.** `--exclude-dynamic-system-prompt-sections` was measured with the same
+> harness (`floor.sh`, n=5, Sonnet 5, on top of `--effort low`): **TTFT median 6.39 s**
+> (5.87–10.54 s) vs **6.17 s** (5.87–6.44 s) for `--effort low` alone — i.e. **0.22 s worse, inside
+> the noise band**, with one worse outlier; dropping that outlier does not change the verdict. n=5
+> cannot prove "zero", only "no benefit detectable above noise" — but there is also a **mechanistic**
+> reason not to expect one: `--help` says the flag *"Improves cross-user prompt-cache **reuse**"*, and
+> **OCP is single-user** — there is no cross-user cache to share, so the flag has nothing to buy here.
+> The banner stayed on `· Claude Max` (no billing-pool drop), but there is no win to bank. The ~6 s
+> floor stands as stated below. Raw rows: [`prefill-spike-measurements.jsonl`](prefill-spike-measurements.jsonl).
+
+
+After #1–#3, the floor is **~6 s**, and it does not go lower. `claude` always injects the full
+Claude Code system prompt + tool definitions (thousands to tens of thousands of prefill tokens)
+regardless of what you ask it. `--exclude-dynamic-system-prompt-sections` exists and may shave
+some of it — **unmeasured**; worth one spike, but do not expect to reach the direct API's
+~1 s.
+
+**Consequence to accept, and to state in the README**: even fully optimized, TUI mode has a
+**~6 s TTFT floor**, so it cannot serve real-time / interactive-latency consumers. It remains
+appropriate for batch, background, and cost-insensitive-latency use. The 知音 AI project
+excluded it on this basis (their prompt-latency budget is 2–4 s) *independently* of the ToS
+question already documented in the README.
+
+---
+
+## Reproduction
+
+```bash
+# harness never touches OCP's :3456 service or ocp-tui-* sessions, and never kill-server
+bash docs/plans/2026-07-13-tui-latency/floor.sh 5                                  # baseline
+TAG=effort-low EXTRA_ARGS="--effort low" bash .../floor.sh 5                       # −40 %
+TAG=bare       EXTRA_ARGS="--bare"       bash .../floor.sh 5                       # the trap
+
+# billing-pool check for ANY spawn-flag change — the banner is the only source of truth
+tmux new-session -d -s probe -x 200 -y 50 -c "$HOME" \
+  "claude --model claude-sonnet-5 --session-id $(uuidgen) <your-flags-here>"
+sleep 6; tmux capture-pane -p -t probe | grep -E "Claude Max|API Usage Billing"
+tmux kill-session -t probe
+```
+
+## Interaction with OCP while the harness runs
+
+- **Kill direction is safe both ways**: `reapStaleTuiSessions()` only `kill-session`s names
+  matching `ocp-tui-<port>-`, which `zhiyin-floor-*` never matches; and the harness only
+  `kill-session`s its own single session — it contains **no `kill-server`**.
+- **One benign interaction** (only when TUI mode is enabled — the reap tick is itself gated on
+  `TUI_MODE`): OCP's periodic `kill-server` (zombie reaping) is gated on
+  `othersRemain` — *any* foreign-prefixed tmux session suppresses it. So while the harness is
+  running, that sweep is skipped. This is the coexistence guard working as designed; it resumes
+  on the next tick.
+
+## Harness caveats (stated so the numbers are not over-trusted)
+
+- **n=5 per config**, single host, single model (Sonnet 5), single prompt size (~1850 tokens).
+  Enough to separate 6 s from 10 s from 30 s; **not** enough for a p95.
+- TTFT is "marker visible in `capture-pane`", which includes tmux render latency (small, but
+  nonzero) — it is an upper bound on the true first-token time.
+- **The harness's readiness marker is not OCP's.** `floor.sh` waits for `│ >|❯|Try "`; OCP's
+  `tuiInputReady()` matches `/\? for shortcuts/`. These are different events, so the ~1.0 s
+  boot figure is **not** directly comparable to OCP's `BOOT_MS` gate (default cap 4000 ms). It
+  does not affect the conclusions (1 s ≪ 6 s TTFT), but it is not apples-to-apples.
+- The first version of this harness reported TTFT **0.08 s** — a false positive: the prompt
+  literally contained the marker string it was grepping for, so the match fired the instant the
+  prompt was pasted. Fixed by describing the marker instead of spelling it. **The script exited 0
+  and "successfully" produced 5 samples both times** — exit status proves nothing here.

docs/plans/2026-07-13-tui-latency/billing-banner.txt

@@ -0,0 +1,3 @@
+[]                     | ▝▜█████▛▘  Sonnet 5 with xhigh effort · Claude Max
+[--effort low]         | ▝▜█████▛▘  Sonnet 5 with low effort · Claude Max
+[--bare]               | ▝▜█████▛▘  Sonnet 5 with xhigh effort · API Usage Billing

docs/plans/2026-07-13-tui-latency/floor.sh

@@ -0,0 +1,128 @@
+#!/usr/bin/env bash
+# OCP TUI-mode latency floor harness — see README.md in this directory.
+#
+# 目的：回答一个问题——如果把 OCP 现有的两个已知开销砍掉
+#   (a) 每请求 spawn + boot（可用预热进程池消除）
+#   (b) 假流式（等 turn_duration 才返回，可用增量读 pane 消除）
+# 之后，订阅池路径的**真实 TTFT 地板**是多少？
+#
+# 判据：地板 ≤ 4s → OCP 作为"省钱选项"可行；> 8s → 死透，不再讨论。
+#
+# 红线：
+# - 不经过生产 OCP 服务（:3456）—— 直接起 tmux+claude，OCP 进程零干扰
+# - tmux session 前缀用 zhiyin-floor-（**不是** ocp-tui-），避免被 OCP 的
+#   reaper 当成自己的会话杀掉，也避免我们杀到它的
+# - 用 real HOME（凭据）—— scratch HOME + symlink 凭据会 fork OAuth 导致 401
+#   （见跨机记忆 tui_scratch_home_credential_fork）
+set -uo pipefail
+
+N=${1:-5}
+MODEL=${MODEL:-claude-sonnet-5}
+EXTRA_ARGS=${EXTRA_ARGS:-}          # 额外 CLI 参数（如 --effort low --bare）
+TAG=${TAG:-baseline}
+OUT=${OUT:-$(dirname "$0")/measurements.jsonl}
+PROMPT_FILE=$(mktemp)
+PREFIX="zhiyin-floor"
+
+mkdir -p "$(dirname "$OUT")"
+
+# ── 构造提示：~2000 token 的假会议转写 + 明确的起始标记 ────────────────
+# 单行（多行会在 tmux send-keys 时提前触发 Enter）
+build_prompt() {
+  local seg="Speaker A said the quarterly pipeline is tracking behind plan and the enterprise segment needs a different motion. Speaker B replied that the current onboarding flow loses roughly a third of trial accounts before the first integration is complete. They debated whether the fix belongs in product or in customer success. "
+  local body=""
+  for _ in $(seq 1 22); do body+="$seg"; done
+  printf '%s' "You are a real-time meeting copilot. Meeting transcript so far: $body --- Task: produce ONE prompt card as compact JSON with keys: points (array of 3 short Chinese bullet points), keyline (one English sentence the user can read aloud). IMPORTANT: your reply MUST begin with three hash characters immediately followed by the uppercase word CARD (no space between them), then the JSON. No preamble, no markdown fences." > "$PROMPT_FILE"
+}
+build_prompt
+PROMPT_CHARS=$(wc -c < "$PROMPT_FILE" | tr -d ' ')
+
+now_ms() { python3 -c 'import time;print(int(time.time()*1000))'; }
+
+echo "配置: $TAG   参数: [$EXTRA_ARGS]"
+echo "模型: $MODEL   样本: $N   提示长度: ${PROMPT_CHARS} chars (≈$((PROMPT_CHARS/4)) token)"
+echo "输出: $OUT"
+echo
+
+for i in $(seq 1 "$N"); do
+  SESS="${PREFIX}-$$-$i"
+  SID=$(uuidgen)
+
+  # ── 冷启动：spawn + 等输入框就绪 ─────────────────────────────────
+  T_SPAWN=$(now_ms)
+  tmux new-session -d -s "$SESS" -x 200 -y 50 \
+    -e CLAUDE_CODE_DISABLE_CLAUDE_MDS=1 \
+    -e CLAUDE_CODE_DISABLE_AUTO_MEMORY=1 \
+    -c "$HOME" \
+    "claude --model $MODEL --session-id $SID --strict-mcp-config --disallowedTools 'mcp__*' $EXTRA_ARGS" 2>/dev/null
+  if [ $? -ne 0 ]; then echo "[$i] tmux spawn 失败，跳过"; continue; fi
+
+  # 轮询输入框就绪（claude TUI 的输入提示符）
+  READY=0
+  for _ in $(seq 1 150); do   # 上限 15s
+    PANE=$(tmux capture-pane -p -t "$SESS" 2>/dev/null || true)
+    if grep -qE '│ >|❯|Try "' <<<"$PANE"; then READY=1; break; fi
+    sleep 0.1
+  done
+  T_READY=$(now_ms)
+  BOOT_MS=$((T_READY - T_SPAWN))
+  if [ "$READY" -ne 1 ]; then
+    echo "[$i] 启动超时（${BOOT_MS}ms），pane 末 3 行:"
+    tmux capture-pane -p -t "$SESS" 2>/dev/null | tail -3 | sed 's/^/      /'
+    tmux kill-session -t "$SESS" 2>/dev/null
+    continue
+  fi
+
+  # ── 热态：粘提示 → 回车 → 量首 token ─────────────────────────────
+  tmux send-keys -t "$SESS" -l "$(cat "$PROMPT_FILE")" 2>/dev/null
+  sleep 0.4                       # 让粘贴落地（OCP 用 400ms 轮询粒度）
+  T0=$(now_ms)
+  tmux send-keys -t "$SESS" Enter 2>/dev/null
+
+  TTFT_MS=-1
+  for _ in $(seq 1 600); do       # 上限 60s
+    if tmux capture-pane -p -t "$SESS" 2>/dev/null | grep -q '###CARD'; then
+      TTFT_MS=$(( $(now_ms) - T0 )); break
+    fi
+    sleep 0.1
+  done
+
+  # ── 完整回答：pane 连续 2s 不再变化 ──────────────────────────────
+  COMPLETE_MS=-1
+  if [ "$TTFT_MS" -ge 0 ]; then
+    LAST=""; STABLE=0
+    for _ in $(seq 1 900); do     # 上限 90s
+      CUR=$(tmux capture-pane -p -t "$SESS" 2>/dev/null | cksum)
+      if [ "$CUR" = "$LAST" ]; then
+        STABLE=$((STABLE+1))
+        [ "$STABLE" -ge 20 ] && { COMPLETE_MS=$(( $(now_ms) - T0 - 2000 )); break; }
+      else
+        STABLE=0; LAST="$CUR"
+      fi
+      sleep 0.1
+    done
+  fi
+
+  printf '{"i":%d,"tag":"%s","model":"%s","extra_args":"%s","prompt_chars":%s,"boot_ms":%d,"ttft_ms":%d,"complete_ms":%d}\n' \
+    "$i" "$TAG" "$MODEL" "$EXTRA_ARGS" "$PROMPT_CHARS" "$BOOT_MS" "$TTFT_MS" "$COMPLETE_MS" | tee -a "$OUT"
+
+  tmux kill-session -t "$SESS" 2>/dev/null
+  sleep 1
+done
+
+rm -f "$PROMPT_FILE"
+echo
+echo "=== 汇总 ==="
+python3 - "$OUT" <<'EOF'
+import json,sys,statistics
+rows=[json.loads(l) for l in open(sys.argv[1]) if l.strip()]
+ok=[r for r in rows if r['ttft_ms']>=0]
+if not ok: print("无有效样本"); sys.exit()
+def s(k):
+    v=[r[k] for r in ok if r[k]>=0]
+    return f"n={len(v)} 中位={statistics.median(v)/1000:.2f}s 最小={min(v)/1000:.2f}s 最大={max(v)/1000:.2f}s" if v else "无"
+print(f"  冷启动 boot      : {s('boot_ms')}   ← 预热进程池可完全消除")
+print(f"  TTFT（首 token） : {s('ttft_ms')}   ★ 这就是地板")
+print(f"  完整回答         : {s('complete_ms')}")
+print(f"\n  失败样本: {len(rows)-len(ok)}/{len(rows)}")
+EOF

docs/plans/2026-07-13-tui-latency/measurements.jsonl

@@ -0,0 +1,15 @@
+{"i": 1, "tag": "effort-low", "model": "claude-sonnet-5", "extra_args": "--effort low", "prompt_chars": 7451, "boot_ms": 1077, "ttft_ms": 6172, "complete_ms": 9929}
+{"i": 2, "tag": "effort-low", "model": "claude-sonnet-5", "extra_args": "--effort low", "prompt_chars": 7451, "boot_ms": 1026, "ttft_ms": 6160, "complete_ms": 9996}
+{"i": 3, "tag": "effort-low", "model": "claude-sonnet-5", "extra_args": "--effort low", "prompt_chars": 7451, "boot_ms": 1010, "ttft_ms": 6437, "complete_ms": 9977}
+{"i": 4, "tag": "effort-low", "model": "claude-sonnet-5", "extra_args": "--effort low", "prompt_chars": 7451, "boot_ms": 1033, "ttft_ms": 5872, "complete_ms": 9944}
+{"i": 5, "tag": "effort-low", "model": "claude-sonnet-5", "extra_args": "--effort low", "prompt_chars": 7451, "boot_ms": 1154, "ttft_ms": 6387, "complete_ms": 9993}
+{"i":1,"tag":"baseline","model":"claude-sonnet-5","extra_args":"","prompt_chars":7451,"boot_ms":1300,"ttft_ms":8321,"complete_ms":9939}
+{"i":2,"tag":"baseline","model":"claude-sonnet-5","extra_args":"","prompt_chars":7451,"boot_ms":1070,"ttft_ms":10347,"complete_ms":11320}
+{"i":3,"tag":"baseline","model":"claude-sonnet-5","extra_args":"","prompt_chars":7451,"boot_ms":911,"ttft_ms":13061,"complete_ms":15163}
+{"i":4,"tag":"baseline","model":"claude-sonnet-5","extra_args":"","prompt_chars":7451,"boot_ms":1441,"ttft_ms":9981,"complete_ms":11066}
+{"i":5,"tag":"baseline","model":"claude-sonnet-5","extra_args":"","prompt_chars":7451,"boot_ms":1036,"ttft_ms":17189,"complete_ms":17985}
+{"i":1,"tag":"bare","model":"claude-sonnet-5","extra_args":"--bare","prompt_chars":7451,"boot_ms":429,"ttft_ms":-1,"complete_ms":-1}
+{"i":2,"tag":"bare","model":"claude-sonnet-5","extra_args":"--bare","prompt_chars":7451,"boot_ms":437,"ttft_ms":-1,"complete_ms":-1}
+{"i":3,"tag":"bare","model":"claude-sonnet-5","extra_args":"--bare","prompt_chars":7451,"boot_ms":444,"ttft_ms":-1,"complete_ms":-1}
+{"i":4,"tag":"bare","model":"claude-sonnet-5","extra_args":"--bare","prompt_chars":7451,"boot_ms":446,"ttft_ms":-1,"complete_ms":-1}
+{"i":5,"tag":"bare","model":"claude-sonnet-5","extra_args":"--bare","prompt_chars":7451,"boot_ms":441,"ttft_ms":-1,"complete_ms":-1}

docs/plans/2026-07-13-tui-latency/messagedisplay-deltas.jsonl

@@ -0,0 +1,7 @@
+{"hook_event_name": "MessageDisplay", "index": 0, "final": false, "delta": "## Mutex\n\n"}
+{"hook_event_name": "MessageDisplay", "index": 1, "final": false, "delta": "A **mutual exclusion lock** prevents concurrent access to a shared resource, ensuring only one thread runs the critical section at a time.\n\n"}
+{"hook_event_name": "MessageDisplay", "index": 2, "final": false, "delta": "- Acquiring a locked mutex blocks the caller until the current holder releases it.\n"}
+{"hook_event_name": "MessageDisplay", "index": 3, "final": false, "delta": "- Failing to release a mutex causes a deadlock, freezing all waiting threads.\n\n```javascript\nconst { Mutex } = require('async-mutex');\n\nconst mutex = new Mutex();\n"}
+{"hook_event_name": "MessageDisplay", "index": 4, "final": false, "delta": "let counter = 0;\n\nasync function increment() {\n  const release = await mutex.acquire();\n  try {\n"}
+{"hook_event_name": "MessageDisplay", "index": 5, "final": false, "delta": "    counter++; // only one caller here at a time\n  } finally {\n    release();\n  }\n}\n"}
+{"hook_event_name": "MessageDisplay", "index": 6, "final": true, "delta": "```"}

docs/plans/2026-07-13-tui-latency/prefill-spike-measurements.jsonl

@@ -0,0 +1,5 @@
+{"i":1,"tag":"effort-low-exclude-dynamic","model":"claude-sonnet-5","extra_args":"--effort low --exclude-dynamic-system-prompt-sections","prompt_chars":7451,"boot_ms":934,"ttft_ms":5867,"complete_ms":9953}
+{"i":2,"tag":"effort-low-exclude-dynamic","model":"claude-sonnet-5","extra_args":"--effort low --exclude-dynamic-system-prompt-sections","prompt_chars":7451,"boot_ms":1275,"ttft_ms":6388,"complete_ms":9874}
+{"i":3,"tag":"effort-low-exclude-dynamic","model":"claude-sonnet-5","extra_args":"--effort low --exclude-dynamic-system-prompt-sections","prompt_chars":7451,"boot_ms":874,"ttft_ms":10537,"complete_ms":11782}
+{"i":4,"tag":"effort-low-exclude-dynamic","model":"claude-sonnet-5","extra_args":"--effort low --exclude-dynamic-system-prompt-sections","prompt_chars":7451,"boot_ms":1170,"ttft_ms":6379,"complete_ms":9947}
+{"i":5,"tag":"effort-low-exclude-dynamic","model":"claude-sonnet-5","extra_args":"--effort low --exclude-dynamic-system-prompt-sections","prompt_chars":7451,"boot_ms":1329,"ttft_ms":6443,"complete_ms":9884}

docs/plans/2026-07-13-tui-latency/streaming-spike.md

@@ -0,0 +1,256 @@
+# Backlog #2 (real streaming): **achievable** — via the `MessageDisplay` hook
+
+**Date**: 2026-07-13
+**Status**: prereq-spike result. **Streaming IS achievable on the TUI path**, byte-faithfully, on the
+subscription pool. Three obvious sources are dead ends; a fourth one works.
+**Scope**: answers the prereq spike that [`README.md`](README.md) § "Backlog #2" demanded *before* any
+streaming design:
+
+> **Prereq spike (do this before designing anything)**: does the transcript JSONL grow *during* a
+> turn, or only at the end? If only at the end, (a) is dead and you are stuck with (b).
+
+The answer: **(a) is dead, (b) is dead — and you are not stuck with either.** The CLI exposes its own
+streaming interface as a **hook**, which the backlog did not consider.
+
+**Measured on**: Mac mini / Claude Code **v2.1.207** / Sonnet 4.6 + Sonnet 5 / Claude Max /
+real-home mode. Every claim below is reproducible from the commands given.
+
+> **Honesty note on how this document was produced.** Its first version concluded the exact opposite —
+> "streaming is not achievable; the CLI exposes no byte-faithful incremental source" — and was **wrong**.
+> An adversarial reviewer, commissioned specifically to *refute* it, found `MessageDisplay` on a second
+> pass; its own first pass had enumerated the hook registry with a truncated grep (it reported 21
+> events — there are **30**). Both the wrong conclusion and its refutation are preserved here, because
+> "we checked, it's impossible" is the most expensive kind of claim to get wrong: it closes a door and
+> nobody re-opens it.
+
+---
+
+## ✅ The source that works: the `MessageDisplay` hook
+
+`claude` fires a **`MessageDisplay`** hook as it renders each block of the assistant's reply. The
+payload carries the **raw markdown source** of an incremental `delta`, plus a monotonic `index` and a
+`final` flag:
+
+```json
+{ "hook_event_name": "MessageDisplay",
+  "turn_id": "6cb31d21-…", "message_id": "84ab9832-…",
+  "index": 0, "final": false, "delta": "## Mutex\n\n" }
+```
+*(payload also carries `session_id`, `transcript_path`, `prompt_id`, `cwd`)*
+
+Registered as an ordinary command hook via `--settings` on a **plain interactive TUI spawn** (no `-p`,
+no `--bare`), `claude-sonnet-4-6`, `--effort low`. Banner verified:
+`▝▜█████▛▘ Sonnet 4.6 with low effort · Claude Max` — **subscription pool, not metered billing**.
+
+One live turn — 7 fires, spread across generation:
+
+```
+index=0 final=false  len=  10  '## Mutex\n\n'
+index=1 final=false  len= 140  'A **mutual exclusion lock** prevents concurrent access to a shar…'
+index=2 final=false  len=  83  '- Acquiring a locked mutex blocks the caller until the current h…'
+index=3 final=false  len= 163  '- Failing to release a mutex causes a deadlock, freezing all wai…'
+index=4 final=false  len=  96  'let counter = 0;\n\nasync function increment() {\n  const release =…'
+index=5 final=false  len=  84  '    counter++; // only one caller here at a time\n  } finally {\n …'
+index=6 final=true   len=   3  '```'
+```
+
+**Every invariant a proxy needs — all hold:**
+
+| requirement | result |
+|---|---|
+| **byte-faithful** — deltas are the model's *source*, not the rendered pane | ✅ `## `, `**`, ```` ```javascript ```` all present in the deltas |
+| **exactness** — `concat(deltas) === T` (the transcript-authoritative text) | ✅ **true**, 579 == 579 bytes |
+| **prefix-stable** — `T.startsWith(concat(deltas[0..n]))` at every n | ✅ **true at all 7 steps** |
+| **incremental** — arrives during generation, not at the end | ✅ 7 fires spread across the turn |
+| **no `-p`** — stays out of the metered `sdk-cli` pool | ✅ plain interactive TUI |
+| **subscription pool** | ✅ banner `· Claude Max` |
+
+This is exactly the contract a streaming design needs: deltas forward straight into SSE
+`delta.content` chunks, and the transcript's final text `T` stays a cheap end-of-turn assertion
+(`concat === T`) instead of a reconciliation problem.
+
+### Caveats for the implementer
+
+- **Block-level granularity, not token-level** — the hook fires **once per rendered block** (roughly one
+  per paragraph / list item / code block), so the chunk count **scales with answer length**: 7 fires for a
+  ~600-byte answer, **18 for a ~2 KB one**. Plenty for SSE (`delta.content` has no minimum size), but do
+  not promise token-by-token output, and do not hard-code any assumption about chunk count.
+- **🔴 The sink MUST be keyed by `session_id` — this is live TODAY, not a future concern.**
+  `OCP_TUI_MAX_CONCURRENT` defaults to **2**, so **two `claude` processes already run concurrently**. One
+  hook command writing to one shared sink would **interleave deltas from two different turns into one
+  stream** — request A's client receiving request B's text, the worst failure a proxy can have, and one a
+  single-request test will never surface. The payload carries `session_id` (and `turn_id` / `message_id`),
+  so demux is easy: derive the sink path from `session_id` (`<dir>/<session_id>.jsonl`) and read only your
+  own turn's file. This *also* keeps the design **warm-pool compatible**, because a pre-booted pane's
+  session-id is fixed at boot — one static hook script serves every pane. **Test it with ≥2 concurrent
+  streaming requests carrying distinguishable prompts and assert zero cross-contamination.**
+- **⚠️ `forceSyncExecution: true` in the hook's source — `claude` BLOCKS on the hook.** A slow hook
+  adds latency to *every* delta. The hook must write and exit immediately (e.g. write to a FIFO / unix
+  socket that OCP reads; never work inline). **Measure the added per-delta latency.**
+- **Thinking blocks appear to be excluded — but this is NOT yet stress-tested. Verify before shipping.**
+  The exclusion is inferred from `content.map(c => c.type === "text" ? c.text : "")` — but that snippet is
+  from the **`final:true`** call site, not the incremental one. Four live turns (incl. two at `--effort
+  high`) showed no thinking text in any delta and `concat === T` held — **but each transcript's thinking
+  block was empty (`thinking:""`, 0 chars)**, so the exclusion was never actually stressed. **The failure
+  mode is severe**: if thinking deltas *do* fire on some config (Opus, `xhigh`), `concat(deltas) !== T`
+  **and OCP streams the model's private reasoning to the caller**. The end-of-turn `concat === T` assertion
+  would *detect* that but **cannot prevent** it — SSE deltas cannot be un-sent. **Before shipping, run a
+  turn on a model+effort that produces substantive thinking** (a hard reasoning prompt on Opus / `xhigh`)
+  and confirm both (a) no thinking text in any delta and (b) `concat === T` still holds.
+- OCP already owns the spawn (isolated HOME, its own flags), so injecting `--settings` with a
+  `MessageDisplay` hook sits inside the existing architecture.
+- **`ALIGNMENT.md`**: this consumes `claude`'s **own** hook surface as emitted — forwarding, not
+  inventing. Not a new endpoint, not a fabricated protocol. (Class B / ADR 0007 — the TUI spawn is
+  OCP-owned; no `cli.js` citation applies.)
+
+### Reproduce in 60 seconds
+
+```bash
+# hook script: append the payload (arrives on stdin) and exit immediately
+printf '#!/bin/bash\ncat >> "$MD_LOG"; printf "\\n" >> "$MD_LOG"; exit 0\n' > /tmp/h.sh && chmod +x /tmp/h.sh
+echo '{"hooks":{"MessageDisplay":[{"hooks":[{"type":"command","command":"MD_LOG=/tmp/deltas.jsonl /tmp/h.sh"}]}]}}' > /tmp/s.json
+
+# plain interactive claude in tmux (prefix NOT ocp-tui-*, and never kill-server)
+tmux new-session -d -s md-probe -x 220 -y 50 \
+  "claude --model claude-sonnet-4-6 --effort low --session-id $(uuidgen) --settings /tmp/s.json"
+# …wait for '? for shortcuts', paste a markdown-producing prompt, press Enter…
+
+jq -r '"\(.index) \(.final) \(.delta|@json)"' /tmp/deltas.jsonl   # incremental raw-markdown deltas
+# then assert: concat(deltas) == extractLatestAssistantText(<transcript>.jsonl)
+```
+
+---
+
+## The three dead ends (still worth knowing — they say what NOT to build)
+
+### (a) Incremental transcript reads — **dead: event granularity, not token granularity**
+
+The transcript JSONL *does* grow during a turn, but one **whole event at a time**; the assistant's text
+event is written as **one complete line**, appearing only ~0.3 s before the terminal `turn_duration`.
+
+Observed (session `efd5b161`, `turn_duration: 7319 ms`):
+
+```
+ #6  t+0.0s   type=user       (the prompt)
+ #15 t+4.7s   type=assistant  blocks=thinking
+ #16 t+7.0s   type=assistant  blocks=text      ← the ENTIRE answer, in one line
+ #21 t+7.3s   type=system     subtype=turn_duration   ← terminal
+```
+
+Cross-checked at **20 ms polling + `fs.watch`** (25× finer): a partial line **never touches disk** —
+one write, `+1` line, carrying the complete answer. Also forced with the undocumented
+`CLAUDE_CODE_INCLUDE_PARTIAL_MESSAGES=1`: still 1 assistant event, 0 partials (interactive mode has no
+stream-json *sink* for it to write to).
+
+**The transcript is still needed** — as the terminal-turn signal, as the authoritative `concat === T`
+check, and as the input to the existing honesty gates (auth-banner detection, `truncated`). It is just
+not the *streaming* source.
+
+### (b) `tmux capture-pane` diffing — **dead: the pane is a RENDERED view, not the text**
+
+The backlog expected to fall back to this, calling it "lossy … (wrapping, scrollback, spinner lines)".
+The loss is far worse than formatting noise: **the pane does not contain the answer's source bytes at
+all.** The TUI *renders* markdown, and `capture-pane -p` strips the ANSI that rendering produced.
+
+Same turn, same lines:
+
+```
+TRANSCRIPT (authoritative T):          PANE (capture-pane -p -J -S -500):
+  '## Semaphore'                         '⏺ Semaphore'                 ← heading marker gone
+  ''                                     ''
+  'A **semaphore** is a synchro…'        '  A semaphore is a synchro…' ← bold markers gone, indented
+```
+
+| token in the answer | in `T` | in the pane's answer region |
+|---|---|---|
+| `## ` (ATX heading) | yes | **no** — rendered as `⏺` |
+| `**` (bold markers) | yes | **no** — rendered to ANSI bold, then stripped by `-p` |
+| ` ```javascript ` (fence + language) | yes | **no** — fence and language tag both gone |
+| `- ` (list item) | yes | yes |
+
+*(A literal `**` does appear elsewhere in the pane — in the **prompt echo**, because the prompt asked
+for bold. Not in the answer.)*
+
+**`capture-pane -e` (keeping the ANSI) does not rescue it — the inverse is provably non-unique.**
+With `T` = ``"## Alpha\n\n**bravo**\n\n```javascript\nlet x=1;\n```"``:
+
+```
+⏺\e[39m \e[1mAlpha\n\n\e[0m  \e[1mbravo\n\n\e[0m  \e[34mlet\e[39m x=\e[32m1\e[39m;
+```
+
+`## Alpha` → **SGR 1 (bold)**. `**bravo**` → **SGR 1 (bold)**. *Identical ANSI* — an H2 and a bold span
+are indistinguishable, never mind `**` vs `__`. The fence and its `javascript` tag are consumed by the
+syntax highlighter into colours; recovering the tag would mean inverting a highlighter, and
+`let x=1;` is valid in several languages.
+
+So `T.startsWith(paneText)` is **false** — raw and indent-stripped, on essentially every markdown
+answer. A proxy streaming pane text would be streaming **something the model did not say**. With
+`MessageDisplay` available there is no reason to go near it.
+
+### (c) `--debug-file` — **dead: it logs stream *timing*, never stream *content***
+
+Worth stating precisely, because a casual check misleads in **both** directions here.
+
+The default log level is `debug`, which **suppresses every `verbose` site**. Raise it and per-chunk
+lines *do* appear, spread across generation:
+
+```bash
+CLAUDE_CODE_DEBUG_LOG_LEVEL=verbose claude --debug-file /tmp/d.log …
+```
+```
+05:51:11.088 [VERBOSE] [shoji-engine] yield stream_event/-     ← 16 of these, mid-turn,
+05:51:11.537 [VERBOSE] [shoji-engine] yield stream_event/-        over ~3.9 s of generation
+05:51:15.192 [DEBUG]   [shoji-engine] turn 1 end (usage in=575 out=255 api=6736ms stop=end_turn resultLen=857)
+```
+
+**But they carry no payload** — the format is `yield <type>/<subtype>`, a bare presence marker. Run with
+no category filter (i.e. all categories) at verbose level: `content_block_delta` = **0**, `text_delta` =
+**0**, `content_block_start` / `message_start` = **0**. The only byte-exact text in the log is the
+end-of-turn `Stop` hook payload (`"last_assistant_message":"## Title\n\n**alpha bravo charlie**"`) —
+transcript granularity. The log tells you **when** tokens arrive, never **what** they are. It is also
+~2.7 MB per turn.
+
+### Also checked, also not the answer
+
+| candidate | outcome |
+|---|---|
+| `--output-format stream-json` (the one interface that emits `text_delta`) | **requires `--print`/`-p`** → `cc_entrypoint=sdk-cli` → the **metered** credit pool, which is exactly what TUI mode exists to avoid. Reproduced live. |
+| `--input-format stream-json` | `Error: --input-format=stream-json requires output-format=stream-json` → same gate. |
+| `CLAUDE_CODE_INCLUDE_PARTIAL_MESSAGES=1` (undocumented) | No stream-json sink in interactive mode → no partials. Banner stayed `· Claude Max`. |
+| `sessionMirror` (undocumented) | Gated on `outputFormat === "stream-json"` → the `-p` family. |
+| `--sdk-url` (hidden) | Forces stream-json + non-interactive → `sdk-cli`. *(inferred from the minified bundle; not banner-tested)* |
+| `~/.claude/sessions/<pid>.json` | Registry metadata only (`{pid, sessionId, cwd, status, version, entrypoint:"cli", kind:"interactive"}`). No assistant text. *(Its `entrypoint:"cli"` incidentally confirms the TUI path stays on the subscription pool.)* |
+| `~/.claude/history.jsonl` | User prompts only; the answer text is absent. |
+| Asking the model to emit plain text (so the pane renders faithfully) | Would mean **mutating the caller's prompt** — a correctness violation for a proxy, and still not byte-faithful (wrapping + indent remain). Rejected. |
+
+---
+
+## Value: what streaming actually buys (read before building)
+
+Streaming is *possible*. Whether it is *worth it* depends on the consumer, and the honest answer is
+uncomfortable:
+
+- **Streaming never makes the answer arrive sooner. It moves the *first* byte, not the *last*.** The
+  final token lands at the same wall-clock moment either way.
+- So a consumer that must have the **complete** answer before it can act — e.g. one parsing a structured
+  JSON reply, **which is exactly the 知音 AI use case that motivated this entire investigation** — gains
+  **nothing at all**. Only a **progressively-rendering** consumer (a chat UI) gains.
+
+And the number the backlog attached to this item was wrong:
+
+- The backlog's "~20 s" was inferred from an external 30–32 s report, **never measured through OCP**.
+  Measured through a real OCP instance (TUI mode, `claude-sonnet-4-6`, ~1850-token prompt, n=5):
+  **median 11.30 s** before [#156](https://github.com/dtzp555-max/ocp/pull/156), **9.55 s** after.
+- **Same-turn decomposition** (baseline row `i=5`): **11.563 s** wall through OCP vs `turn_duration:
+  7.319 s` of CLI-internal time on that same turn → **OCP's own overhead ≈ 4.2 s** (n=1, baseline
+  `effort=high` config). *Caveats*: n=1; and `turn_duration` is the CLI's internal duration of an
+  **OCP-driven** turn, not a separate "native" baseline. Do **not** subtract this `effort=high` 7.3 s
+  from the `effort=low` 9.55 s median — a low-effort turn generates faster, so mixing them
+  *understates* the overhead.
+- So OCP's own overhead is **single-digit seconds**, not ~20 s. The rest of any large number is the
+  model generating a long answer — which streaming hides but does not shorten.
+
+**Recommendation**: build it — the contract is clean and the cost is small — but size the expectation
+honestly. It is a *perceived-latency* feature for progressively-rendering consumers, not a throughput
+win, and it does not move the **~6 s TTFT floor** ([`README.md`](README.md)) that rules TUI mode out for
+interactive-latency consumers regardless.

docs/plans/anthropic-sandbox-strategy.md

@@ -0,0 +1,268 @@
+# OCP Anthropic-Only Sandbox Strategy — Handoff Document
+
+**Status:** Forward-looking planning doc (not yet a decision)
+**Date:** 2026-05-29
+**Audience:** future OCP maintainer / session picking up multi-tenant security work
+**Provenance:** authored during OLP Phase 7 PR-B re-evaluation; OLP's parallel analysis (multi-provider) lives at `dtzp555-max/olp` `docs/adr/0014-sandbox-runtime-integration.md` Amendment 1 (pending). This OCP-side doc strips the multi-LLM generalization and keeps only what applies to OCP's single-provider (anthropic) deployment.
+
+---
+
+## 1. Why this doc exists
+
+OCP is in maintenance mode (per OLP ADR 0001 supersession of OCP ADR 0005). It is not under active development for new features. However, two things may eventually drive sandbox work in OCP:
+
+1. **Multi-key OCP deployments.** `OCP_OWNER_TOKEN` + per-key cache namespace already shipped (OCP `lib/keys.mjs`). If multiple human users share an OCP instance, the same multi-tenant filesystem-isolation gap that motivated OLP Phase 7 also exists here.
+2. **Cloud or shared-host OCP deployments.** Any deployment beyond "single user on their own machine" inherits the threat surface.
+
+If/when that work starts, this doc is the prior-art capture so the maintainer doesn't repeat OLP's PR-B path (which has a documented dead-end — see § 3.2 below).
+
+This doc is anthropic-only by design — codex/mistral/etc. multi-LLM concerns are out of scope per OCP ADR 0005.
+
+---
+
+## 2. The multi-tenant gap (OCP-specific)
+
+OCP spawns `claude -p` as the OCP-process user. Every spawned claude instance runs with the OCP user's filesystem permissions. Consequences for a multi-key OCP deployment:
+
+1. **Cross-key lateral read.** A prompt-injected `cat ~/.ocp/keys/<other-key>.json` reads any other key's manifest (token hash, owner_tier, providers_enabled — not catastrophic since it's only the *hash*, but still identity-attribution surface).
+2. **OAuth credential exposure.** `~/.claude/.credentials.json` is the Anthropic OAuth refresh token. A prompt-injected read of this file = stealing the subscription that OCP exists to pool.
+3. **SSH identity exposure.** `~/.ssh/id_*` reachable for lateral movement to other hosts the OCP user can reach.
+4. **Other host secrets.** Anything else under the OCP user's home is reachable.
+
+OCP's `ALIGNMENT.md` Class A/B endpoint discipline does not address this — that discipline is wire-level honesty (`cli.js` mirror), not host-level isolation.
+
+The threat model assumes prompt-injection capability — any caller with a valid OCP key + ability to craft a prompt that elicits a tool call. Default `claude -p` mode includes Read/Bash/etc. tool descriptions in the system prompt; the model is **eager** to use them.
+
+---
+
+## 3. Why OLP Phase 7 PR-B is the wrong path to copy
+
+OLP attempted to wrap `claude -p` spawn in `@anthropic-ai/sandbox-runtime` (outer bubblewrap on Linux, sandbox-exec on macOS). This produced four binding problems documented during OLP's re-evaluation:
+
+### 3.1 Anthropic's design doesn't expect external sandboxing
+
+Per Anthropic's [engineering blog on Claude Code sandboxing](https://www.anthropic.com/engineering/claude-code-sandboxing), `sandbox-runtime` is designed to be invoked **by claude code itself** to sandbox **its own** Bash tool / MCP servers / spawn children. It is **not** designed to sandbox claude code as an externally-wrapped process.
+
+Concretely: claude CLI assumes it can freely read+write its own `$HOME`-derived paths (`~/.claude.json`, `~/.claude/.credentials.json`, `~/.config/claude/`, future state files). When wrapped in `bwrap --ro-bind / /`, those writes hit `EROFS` and claude silently exits with no stdout.
+
+### 3.2 `~/.claude.json` upstream status is "closed not planned"
+
+claude CLI writes `~/.claude.json` non-atomically at startup. Upstream issues #28842, #29162, #29217, #28837, #29051, #29250, #7243 all document this. **#29250 is closed as "not planned / duplicate"** — Anthropic is not going to make this file atomic-write because their mental model is that claude runs in an environment that can write its `$HOME`.
+
+For OCP, this means: any outer-sandbox approach that uses `--ro-bind` on `$HOME` will be a **permanent maintenance treadmill** — every new claude CLI version that adds a state file outside the patched mount paths breaks OCP. OLP's PR-B fold-in tried to patch this by promoting `~/.claude/` to rw, which was insufficient (the actual file is `~/.claude.json` at $HOME root, not inside `~/.claude/`).
+
+### 3.3 The threat model doesn't justify the cost
+
+OCP is, per ADR 0005, a personal-and-family-scale tool. The realistic threat surface is misbehaving prompts from family members or self-injected via dependent agents, not adversarial external attackers. The blast radius of a successful cross-key read is bounded (token *hash*, OAuth that's pooled-by-design across all OCP keys).
+
+A maintenance-mode project investing weeks into outer-sandboxing for a hypothetical threat is a poor cost/benefit. There are cheaper architectures (§ 4 below) that get most of the protection.
+
+### 3.4 OLP-specific reason that does NOT apply to OCP
+
+OLP also hit a multi-provider conflict: codex CLI has its own inner bubblewrap that breaks when wrapped in an outer bwrap (openai/codex#16018). **This is not an OCP concern** — OCP only spawns claude. So the multi-provider forcing function for OLP doesn't apply here. The other three reasons (§ 3.1–3.3) are sufficient on their own.
+
+---
+
+## 4. Three viable approaches for OCP
+
+Ranked by "engineering cost vs isolation strength" — pick by deployment context.
+
+### 4.1 Approach A — Ephemeral `$HOME` via env var (recommended starting point)
+
+Per-spawn setup:
+
+```
+ephemeralRoot=/tmp/ocp-spawn/<keyId>/<reqId>/home
+mkdir -p $ephemeralRoot/.claude
+ln -s ~/.claude/.credentials.json $ephemeralRoot/.claude/.credentials.json
+HOME=$ephemeralRoot claude -p --output-format stream-json ...
+```
+
+Mechanics:
+- claude CLI uses Node's `os.homedir()` which reads `$HOME` env first.
+- `~/.claude.json` written by claude on startup → lands in `/tmp/ocp-spawn/<keyId>/<reqId>/home/.claude.json` (tmpfs, discarded after spawn).
+- `~/.claude/.credentials.json` is the OAuth file claude needs — symlinked in read-only from the real one.
+- Any new state file claude CLI introduces in a future version → also lands in the ephemeral home, no patch needed.
+
+Threat coverage:
+- ✅ Solves EROFS upgrade tax permanently — any claude state-file location works because they all land in tmpfs.
+- ✅ Cross-key OAuth credential isolation — keyA's ephemeral home has only keyA's symlink, but here the symlink target is the SAME real file because OCP shares OAuth (this is fine: shared OAuth is OCP's design, the symlink just keeps the file inaccessible via `cat ~/.claude/.credentials.json` from a different keyId's ephemeral root).
+- ❌ Does NOT solve cross-key lateral filesystem read via absolute paths. A prompt-injected `cat /home/<ocp-user>/.ocp/keys/<otherKey>.json` still works — `os.homedir()` override doesn't affect absolute-path reads.
+
+5-minute spike before adopting:
+
+```bash
+HOME=/tmp/fake-home-spike claude --print "echo PONG" --no-session-persistence 2>&1
+ls -la /tmp/fake-home-spike  # expect: .claude.json + .claude/ created here
+find ~/.claude ~/.claude.json -newer /tmp/spike-marker 2>/dev/null  # expect: empty
+```
+
+If claude falls back to `os.userInfo().homedir` (uses getpwuid_r, ignores HOME env), this approach degrades — fall back to Approach B.
+
+**Engineering cost:** ~50 LOC in OCP's spawn pipeline (mkdir + symlink + env merge + cleanup-on-exit). No new dependencies.
+
+### 4.2 Approach B — Outer bubblewrap with `--tmpfs $HOME` + `--ro-bind` credentials
+
+```
+bwrap \
+  --ro-bind / / \
+  --tmpfs /home/<ocp-user> \
+  --ro-bind /home/<ocp-user>/.claude/.credentials.json /home/<ocp-user>/.claude/.credentials.json \
+  --ro-bind /home/<ocp-user>/.ocp/keys/<thisKeyId>.json /home/<ocp-user>/.ocp/keys/<thisKeyId>.json \
+  --dev /dev --proc /proc --tmpfs /tmp \
+  claude -p ...
+```
+
+This is the canonical bwrap pattern (Flatpak uses exactly this for every sandboxed app — see [Bubblewrap ArchWiki Examples](https://wiki.archlinux.org/title/Bubblewrap/Examples)).
+
+Threat coverage:
+- ✅ Solves EROFS upgrade tax (tmpfs accepts any write path).
+- ✅ Cross-key lateral read prevention — only the current key's manifest is bind-mounted in, others are simply absent from the sandbox view.
+- ✅ `~/.ssh` and similar identity material absent from sandbox.
+
+Trade-offs:
+- bwrap dependency: install `bubblewrap` apt package on host.
+- Bypasses `@anthropic-ai/sandbox-runtime` library — direct bwrap arg composition. Worth it because sandbox-runtime's outer-wrap design is for short-lived claude-internal subprocesses, not long-running claude CLI itself (per § 3.1).
+- macOS: not supported by bwrap (macOS would need separate `sandbox-exec` profile, ~50-100 LOC additional work). OCP cross-machine maintainer deploys mostly on Mac mini + Oracle ARM VM — both Linux on the cloud side, Mac mini side may remain unsandboxed if family-trust-zone.
+
+**Engineering cost:** ~150 LOC for the spawn wrapper + deployment doc updates to require `apt install bubblewrap`. macOS support is a separate ~100 LOC if/when needed.
+
+### 4.3 Approach C — OverlayFS lowerdir (read-only) + tmpfs upperdir (writable)
+
+```
+mount -t overlay overlay \
+  -o lowerdir=/home/<ocp-user>/.claude,upperdir=/tmp/ocp-spawn/<reqId>/upper,workdir=/tmp/ocp-spawn/<reqId>/work \
+  /tmp/ocp-spawn/<reqId>/merged-claude
+HOME=/tmp/ocp-spawn/<reqId>/home claude -p ...
+# After spawn: umount + rm -rf
+```
+
+Most elegant — claude sees a view identical to its real `~/.claude/`, all writes go to tmpfs upperdir, real `~/.claude/` is never touched.
+
+Trade-offs:
+- Requires `CAP_SYS_ADMIN` or rootless-overlayfs (kernel ≥5.11 + user-ns enabled). OCP currently runs as the maintainer's user — no SYS_ADMIN — so this would require either running OCP as root (bad) or rootless-overlayfs setup.
+- More moving parts (mount/umount per spawn, work-dir lifetime, cleanup-on-crash).
+
+Better fit if OCP ever moves to a dedicated `ocp` system user with `CAP_SYS_ADMIN` capability via systemd.
+
+**Engineering cost:** ~120 LOC + kernel/permission preflight check.
+
+---
+
+## 5. Cross-key isolation orthogonal layer
+
+The three approaches above all solve `~/.claude.json` EROFS + state-write isolation. None of them alone solve **cross-key lateral filesystem read via absolute paths** (e.g. prompt-injected `cat /home/<user>/.ocp/keys/<otherKey>.json`).
+
+For that, two options compose with any of A/B/C:
+
+### 5.1 Per-spawn `sandbox-runtime` customConfig with `denyRead`
+
+`@anthropic-ai/sandbox-runtime`'s `wrapWithSandbox(command, binShell?, customConfig?, abortSignal?)` accepts per-call override:
+
+```
+const otherKeysWorkspaces = listAllKeyManifestsExcept(thisKeyId)
+const wrapped = await SandboxManager.wrapWithSandbox(claudeCommand, undefined, {
+  filesystem: {
+    denyRead: [
+      ...otherKeysWorkspaces,          // all keys except current
+      '/home/<ocp-user>/.ssh',
+      '/home/<ocp-user>/.gnupg',
+      '/home/<ocp-user>/.aws',
+    ],
+    allowWrite: [ephemeralRoot, '/tmp'],
+  },
+})
+```
+
+This adds bwrap deny-paths per-spawn (after sandbox-runtime singleton init). Works in combination with Approach A (the `HOME` env-var override is independent of sandbox-runtime's restrictions).
+
+Caveat: this re-introduces the outer-bwrap concern from § 3.1 — claude CLI is now wrapped after all. Mitigation: use this only for **cross-key isolation**, not for `$HOME` restriction. The `denyRead` paths are all outside `$HOME`, so claude's `~/.claude.json` write is unaffected.
+
+### 5.2 Per-OS-user OCP spawning
+
+Each OCP key gets a dedicated Linux user (`ocp-<keyId>`). Spawn claude as that user via `runuser` or `sudo -u`. OAuth credential shared via Linux group permissions or bind-mount.
+
+True kernel-level uid isolation. Most robust answer for OCP-as-shared-host scenarios.
+
+Trade-offs:
+- Setup script complexity (one-time per key).
+- Linux-only.
+- Doesn't fit Mac mini deployment.
+
+Best fit for a cloud OCP deployment where per-tenant trust isolation matters.
+
+---
+
+## 6. Trust model framing
+
+OCP's authentication layer (`lib/keys.mjs`) provides **attribution** (per-key audit, per-key cache namespace). It does NOT, by itself, provide **isolation** (per-key trust boundary against prompt-injection lateral reads).
+
+This distinction is worth making explicit in OCP's README "Security" section (it currently isn't). The three tiers:
+
+| Tier | Trust Model | Sandbox requirement |
+|---|---|---|
+| **Single-user** | maintainer's own machine, single OCP token | None — system-user permissions are sufficient |
+| **Family-trust-zone** | maintainer + family members on shared OCP instance, all parties trusted not to attack each other | Optional — Approach A (ephemeral $HOME) gives cleanup hygiene without changing trust assumptions |
+| **Shared-host / cloud / external callers** | OCP keys handed to potentially-adversarial callers (CI runners, third-party agents, public demo) | Required — Approach B or C + § 5 cross-key isolation |
+
+The current OCP deployment fits tier 1 or 2. The work in this doc applies only when promoting to tier 3.
+
+---
+
+## 7. Recommendation if/when this work starts
+
+**Phase 1 — Approach A (ephemeral `$HOME`) only.**
+- ~50 LOC, no apt deps, works on Mac mini + Linux
+- Solves the EROFS upgrade tax structurally
+- Closes cross-key OAuth-credential-file lateral read
+- Cost-effective hygiene improvement
+
+**Phase 2 — Approach B (outer bwrap) gated by deployment config.**
+- Add `~/.ocp/config.json` field `security.sandbox: 'off' | 'tmpfs-home'`
+- Default off (preserves Mac mini family deployment)
+- Operator opts in on Linux cloud deployments
+- Apt prereq documented in deployment guide
+
+**Phase 3 — § 5 cross-key isolation (only if tier 3 deployment is planned).**
+- Layer per-spawn customConfig denyRead OR per-OS-user spawning
+- Treat as separate ADR amendment with its own threat-model evidence
+
+**Skip Approach C** unless a future requirement forces overlay (low likelihood for OCP scope).
+
+---
+
+## 8. Authority citations
+
+This doc claims findings about claude CLI / `@anthropic-ai/sandbox-runtime` behavior. Sources for verification:
+
+- [Anthropic engineering — Claude Code sandboxing](https://www.anthropic.com/engineering/claude-code-sandboxing) (sandbox-runtime design intent)
+- [Anthropic sandbox-runtime GitHub](https://github.com/anthropic-experimental/sandbox-runtime) (wrapWithSandbox API + customConfig per-call signature)
+- [claude-code#29250 — `.claude.json` non-atomic-write closed-not-planned](https://github.com/anthropics/claude-code/issues/29250)
+- [claude-code#29162 — read-only `~/.claude.json` startup hang](https://github.com/anthropics/claude-code/issues/29162)
+- [claude-code#29217 — concurrent-write corruption](https://github.com/anthropics/claude-code/issues/29217)
+- [claude-code#28842 — Windows startup race](https://github.com/anthropics/claude-code/issues/28842)
+- [claude-code#7243 — "the .claude.json elephant in the room"](https://github.com/anthropics/claude-code/issues/7243)
+- [Bubblewrap README](https://github.com/containers/bubblewrap)
+- [Bubblewrap ArchWiki — Examples section, --tmpfs HOME pattern](https://wiki.archlinux.org/title/Bubblewrap/Examples)
+- [Sandboxing CLI tools with Bubblewrap — botmonster](https://botmonster.com/self-hosting/sandbox-linux-apps-cli-tools-bubblewrap/)
+- [OverlayFS kernel documentation](https://docs.kernel.org/filesystems/overlayfs.html)
+- [OverlayFS ArchWiki](https://wiki.archlinux.org/title/Overlay_filesystem)
+
+OLP's parallel work (multi-provider generalization of this strategy, including the codex inner-bwrap conflict that does not apply to OCP):
+
+- `dtzp555-max/olp` `docs/adr/0014-sandbox-runtime-integration.md` (PR-B as-shipped) + Amendment 1 (pending — Solution 1 architecture)
+- `dtzp555-max/olp` `docs/plans/cloud-deployment-family.md` § 5 (deployment-side trust tier mapping)
+- archive branch `dtzp555-max/olp:phase-7-pr-b-outer-bwrap-snapshot` captures the outer-bwrap approach as snapshot if anyone wants to revisit it
+
+---
+
+## 9. What this doc is NOT
+
+- Not an ADR. ADRs are decisions; this is a forward-facing strategy doc that becomes an ADR only when work starts and a decision is made.
+- Not a binding spec. The three approaches are alternatives; the recommendation in § 7 is the maintainer's lean from prior-art analysis, not a constitution.
+- Not authority for any code change. OCP `ALIGNMENT.md` still requires citation per Class A/B; no sandbox code lands without proper authority pinning when the work eventually starts.
+- Not a security audit. The threat model is informal — based on prior-art search + incident memory from OLP's parallel session. A real cloud deployment should commission an independent threat model.
+
+---
+
+**Authors:** project maintainer (handoff prepared with AI drafting assistance during OLP Phase 7 PR-B re-evaluation, 2026-05-29).

docs/runbooks/615-canary.md

@@ -0,0 +1,151 @@
+# 2026-06-15 Canary Runbook
+
+**Purpose:** Confirm that a TUI-mode turn is billed to the **Pro/Max subscription pool** (not the Agent SDK credit pool) after Anthropic's 2026-06-15 billing split activates.
+
+The billing classifier reading `cli` is **necessary but NOT sufficient** proof. (Note the naming: the value is stored in the JSONL transcript under the field name `entrypoint`, and sent to Anthropic on the wire as the `cc_entrypoint` header — they carry the same value after claude's startup classification. The commands below grep the transcript, so they match `entrypoint`.) A `cli` label tells you OCP sent the right classification; it does not tell you Anthropic billed the right pool. The only authoritative test is to observe whether the **Agent SDK credit balance** moves or not before and after the canary turn.
+
+---
+
+## Prerequisites
+
+- `CLAUDE_TUI_MODE=true` already set and OCP restarted (see [TUI-mode setup](../tui-mode.md#enabling-tui-mode-opt-in))
+- `tmux` installed on the host
+- No other OCP traffic during the canary (quiesce — see below)
+- Access to your Anthropic account billing page (manual step — see below)
+
+---
+
+## Step 1 — Quiesce the host
+
+Stop any IDE or client that is actively sending requests through this OCP instance.
+
+Confirm the proxy is idle:
+
+```bash
+curl -s http://127.0.0.1:3456/health | python3 -m json.tool | grep activeRequests
+# Expected: "activeRequests": 0
+```
+
+Wait until `activeRequests` is `0` before proceeding. If you cannot quiesce (e.g. family members are actively using it), run the canary on a separate OCP instance or during a quiet window.
+
+---
+
+## Step 2 — Read the Agent SDK credit balance BEFORE the canary
+
+> **Manual step — no programmatic API available.**
+>
+> OCP's `/usage` endpoint reads `anthropic-ratelimit-unified-*` response headers from the Pro/Max plan quota (5-hour and 7-day subscription windows). These headers report **subscription usage**, not the Agent SDK credit pool balance. There is no known programmatic API to query the Agent SDK credit pool balance from outside the Anthropic web app.
+
+To read the balance:
+
+1. Open [https://claude.ai/settings/billing](https://claude.ai/settings/billing) (or your Anthropic Console billing page) in a browser.
+2. Find the **Agent SDK Credits** section (sometimes labeled "API Credits" or "Agent SDK usage").
+3. Note the current balance (e.g. `$18.43 remaining of $20.00`).
+
+Write the value down — you will compare it after the canary turn.
+
+---
+
+## Step 3 — Send the canary turn
+
+With TUI-mode on and the host quiesced, send exactly one small request:
+
+```bash
+curl -s -X POST http://127.0.0.1:3456/v1/chat/completions \
+  -H "Content-Type: application/json" \
+  -d '{
+    "model": "claude-haiku-4-5-20251001",
+    "messages": [{"role": "user", "content": "Reply with the single word: pong"}],
+    "max_tokens": 10
+  }' | python3 -m json.tool
+```
+
+Use Haiku (the cheapest model) to minimize any hypothetical impact if the canary turns red.
+
+Wait for the response to arrive completely (TUI-mode buffers the full response before returning — you will see a delay of several seconds, then the full reply).
+
+---
+
+## Step 4 — Confirm the transcript shows `entrypoint:"cli"`
+
+After the canary turn completes, inspect the most recent JSONL transcript for the billing-classifier label:
+
+```bash
+# The canary was run quiesced (Step 1), so the most recent JSONL across ALL project
+# dirs IS the canary turn. We glob every projects subdir instead of recomputing
+# claude's cwd-encoding rule (it maps every "/" AND "." to "-", e.g. ~/.ocp-tui/work
+# => projects/-home-<user>--ocp-tui-work/; see lib/tui/transcript.mjs encodeCwd) —
+# a glob is robust even if that encoding changes in a future claude build.
+LATEST=$(ls -t "$HOME"/.claude/projects/*/*.jsonl 2>/dev/null | head -1)
+echo "Transcript: $LATEST"
+grep -o '"entrypoint":"[^"]*"' "$LATEST" | tail -1
+# Expected: "entrypoint":"cli"
+```
+
+If the output shows `"entrypoint":"cli"`, the billing-classifier label is correct. If it shows `"entrypoint":"sdk-cli"`, the spawn did not get a real PTY — stop immediately and do not re-enable TUI-mode without investigation. Check `tmux new-session` manually and review ADR 0007 § spawn/PTY gate. (If the grep returns nothing, the transcript may not yet be flushed — re-run after a second, or confirm the turn completed.)
+
+**Reminder: an `entrypoint:cli` label (the `cc_entrypoint=cli` wire header) is necessary but not sufficient.** It tells you OCP sent the right label to Anthropic. You must still check the credit balance in Step 5.
+
+---
+
+## Step 5 — Re-read the Agent SDK credit balance AFTER the canary
+
+Return to [https://claude.ai/settings/billing](https://claude.ai/settings/billing) and reload the page. Note the current balance again.
+
+---
+
+## Step 6 — Green/Red decision
+
+### Green (balance unchanged)
+
+The Agent SDK credit balance did not decrease. The turn billed against the Pro/Max subscription pool as expected. TUI-mode is working correctly.
+
+**Actions:**
+- Keep `CLAUDE_TUI_MODE=true` on this host.
+- Monitor the balance periodically for the first week to catch any delayed attribution.
+- Resume normal traffic.
+
+### Red (Agent SDK credit balance decreased)
+
+The Agent SDK credit balance decreased. The subscription pool is not being used for TUI-mode turns on this host, despite `cc_entrypoint=cli` being set. This may indicate a backend routing change on Anthropic's side, a TTY detection failure, or a policy change.
+
+**Actions — immediate:**
+1. Unset `CLAUDE_TUI_MODE` (or set to any value other than `"true"`) in the service unit:
+   - systemd: edit `/etc/ocp/ocp.env` (or the unit's `Environment=` line), then `sudo systemctl daemon-reload && sudo systemctl restart ocp.service`
+   - launchd: edit the plist `EnvironmentVariables` section, then `launchctl bootout gui/$(id -u)/dev.ocp.proxy && launchctl bootstrap gui/$(id -u) <plist-path>`
+2. Restart OCP and confirm the `/health` response no longer shows TUI-mode active.
+3. If you share this OCP with family or other Max users: freeze their access temporarily until you understand the billing impact.
+4. Consider pivoting to OLP multi-provider (see [OLP](https://github.com/dtzp555-max/olp)) which can spread load across other providers to avoid the Agent SDK credit drain.
+
+Per ALIGNMENT.md Rule 2 / ADR 0007 § Kill-switch: "Per the constitution, the response is to drop the Anthropic provider rather than escalate spoofing."
+
+---
+
+## Ongoing monitoring — self-classification mini-canary
+
+To detect future drift (e.g. a claude CLI upgrade that changes TTY-detection behavior), you can run a periodic one-liner that sends a tiny TUI turn with `OCP_TUI_ENTRYPOINT=auto` (so claude self-classifies rather than having OCP pin the value) and alerts if the transcript self-classification is not `cli`:
+
+```bash
+# Run with OCP temporarily configured OCP_TUI_ENTRYPOINT=auto
+# Then check the most recent transcript:
+# Glob the most recent transcript across all project dirs (robust to claude's
+# cwd-encoding rule; run this right after the auto-mode mini-canary turn).
+LATEST=$(ls -t "$HOME"/.claude/projects/*/*.jsonl 2>/dev/null | head -1)
+RESULT=$(grep -o '"entrypoint":"[^"]*"' "$LATEST" | tail -1)
+echo "Self-classified entrypoint: $RESULT"
+if echo "$RESULT" | grep -q '"entrypoint":"cli"'; then
+  echo "OK — subscription pool"
+else
+  echo "ALERT — not cli; check TTY and billing"
+fi
+```
+
+Run this after any major `claude` CLI upgrade. The `auto` mode lets the CLI's own `t$A` startup function determine the value from the actual TTY state (see ADR 0007 § Billing-classifier labeling).
+
+---
+
+## Related
+
+- [Flip/rollback runbook](./tui-flip-rollback.md) — how to set and unset `CLAUDE_TUI_MODE` on systemd and launchd hosts
+- [ADR 0007](../adr/0007-tui-interactive-mode.md) — TUI-mode architecture and governing rules
+- [Subscription-pool (TUI) mode](../tui-mode.md#subscription-pool-tui-mode)

docs/runbooks/tui-flip-rollback.md

@@ -0,0 +1,180 @@
+# TUI-Mode Flip and Rollback Runbook
+
+**Purpose:** Step-by-step instructions for enabling (`CLAUDE_TUI_MODE=true`) or disabling TUI-mode on real OCP deployments managed by **systemd** (Linux) or **launchd** (macOS).
+
+Run the [615-canary](./615-canary.md) runbook after any flip to confirm billing pool routing is correct.
+
+---
+
+## Critical pitfalls — read first
+
+### systemd: `daemon-reload` is required after editing the unit
+
+Editing the unit file (or EnvironmentFile) and then doing `systemctl restart ocp.service` **without** `daemon-reload` will restart the process with the **old** environment from the cached unit. Always run `daemon-reload` after editing any unit file.
+
+### launchd: `launchctl kickstart -k` does NOT reload plist env
+
+`launchctl kickstart -k gui/$(id -u)/dev.ocp.proxy` kills the running process and re-launches it, but it **re-uses the launchd-cached environment** — not the current plist file. If you edited the plist's `EnvironmentVariables` section, you must do a full `bootout` + `bootstrap` cycle for the change to take effect. `kickstart` is not sufficient.
+
+---
+
+## Flip — enable TUI-mode
+
+### systemd (Linux, e.g. Raspberry Pi, VPS)
+
+**Option A — EnvironmentFile (recommended for clean separation)**
+
+If your unit uses `EnvironmentFile=/etc/ocp/ocp.env` (or similar):
+
+```bash
+# 1. Edit the environment file
+sudo nano /etc/ocp/ocp.env
+# Add or update:
+#   CLAUDE_TUI_MODE=true
+#
+# If OCP binds to 0.0.0.0 AND you trust the network:
+#   OCP_TUI_ALLOW_LAN=1
+#   (WARNING: TUI-mode is single-user only — only enable OCP_TUI_ALLOW_LAN=1
+#    if you fully trust every caller that can reach the OCP port on your network)
+
+# 2. Reload the unit definition and restart
+sudo systemctl daemon-reload
+sudo systemctl restart ocp.service
+
+# 3. Verify
+curl -s http://127.0.0.1:3456/health | python3 -m json.tool | grep -E "tui|version"
+# Expected: "tuiMode": true (or similar TUI indicator in the health response)
+```
+
+**Option B — inline Environment= in the unit file**
+
+```bash
+# 1. Edit the unit file
+sudo systemctl edit --full ocp.service
+# Add or update in the [Service] section:
+#   Environment=CLAUDE_TUI_MODE=true
+
+# 2. Reload and restart
+sudo systemctl daemon-reload
+sudo systemctl restart ocp.service
+
+# 3. Verify
+systemctl show ocp.service --property=Environment
+# Expected: Environment=CLAUDE_TUI_MODE=true ...
+```
+
+### launchd (macOS)
+
+Locate the OCP plist. The standard label is `dev.ocp.proxy`:
+
+```bash
+# Find the plist path
+ls ~/Library/LaunchAgents/dev.ocp.proxy.plist
+```
+
+**Edit the plist:**
+
+```bash
+# 1. Stop the service first (bootout)
+launchctl bootout gui/$(id -u)/dev.ocp.proxy
+
+# 2. Edit the plist — add CLAUDE_TUI_MODE to EnvironmentVariables
+#    Use your editor of choice:
+nano ~/Library/LaunchAgents/dev.ocp.proxy.plist
+```
+
+Inside the plist, in the `<key>EnvironmentVariables</key>` `<dict>` block, add:
+
+```xml
+<key>CLAUDE_TUI_MODE</key>
+<string>true</string>
+```
+
+If `OCP_TUI_ALLOW_LAN=1` is also needed (only if OCP binds to `0.0.0.0` and you trust the network):
+
+```xml
+<key>OCP_TUI_ALLOW_LAN</key>
+<string>1</string>
+```
+
+```bash
+# 3. Bootstrap (reload from disk + start)
+launchctl bootstrap gui/$(id -u) ~/Library/LaunchAgents/dev.ocp.proxy.plist
+
+# 4. Verify
+curl -s http://127.0.0.1:3456/health | python3 -m json.tool | grep -E "tui|version"
+```
+
+**Confirm env was actually loaded** (not just set in your shell):
+
+```bash
+ps aux | grep server.mjs | grep -v grep
+# Get the PID, then:
+# macOS: ps -E -p <PID> | tr ' ' '\n' | grep CLAUDE_TUI_MODE
+# Expected: CLAUDE_TUI_MODE=true
+```
+
+---
+
+## Rollback — disable TUI-mode
+
+Rollback is the same procedure as flip, but you **remove** `CLAUDE_TUI_MODE` or set it to any value other than `"true"` (e.g. `false`, or simply omit it).
+
+After rollback, OCP returns to the default `callClaude` / `callClaudeStreaming` stream-json path — byte-for-byte identical to the pre-TUI code path. No other change is required.
+
+### systemd rollback
+
+```bash
+# Option A — EnvironmentFile
+sudo nano /etc/ocp/ocp.env
+# Remove or comment out:
+#   CLAUDE_TUI_MODE=true
+#   OCP_TUI_ALLOW_LAN=1  (if set)
+
+sudo systemctl daemon-reload
+sudo systemctl restart ocp.service
+
+# Verify
+curl -s http://127.0.0.1:3456/health | python3 -m json.tool | grep tui
+# Expected: "tuiMode": false (or the field absent)
+```
+
+### launchd rollback
+
+```bash
+# 1. Stop
+launchctl bootout gui/$(id -u)/dev.ocp.proxy
+
+# 2. Edit plist — remove the CLAUDE_TUI_MODE and OCP_TUI_ALLOW_LAN entries from EnvironmentVariables
+
+# 3. Bootstrap
+launchctl bootstrap gui/$(id -u) ~/Library/LaunchAgents/dev.ocp.proxy.plist
+
+# 4. Verify
+curl -s http://127.0.0.1:3456/health | python3 -m json.tool | grep tui
+```
+
+---
+
+## Billing impact of staying on the default (non-TUI) path after 2026-06-15
+
+If you do NOT flip to TUI-mode and keep `CLAUDE_TUI_MODE` unset (the default), OCP continues using `claude -p --output-format stream-json`, which sets `cc_entrypoint=sdk-cli`. After 2026-06-15, every OCP request on the default path will draw from the Agent SDK credit pool (approximately $20/month on a Pro plan, or $100/month on a Max plan) rather than the Pro/Max subscription. The subscription pool usage (5-hour and 7-day windows) will be unaffected, but the Agent SDK credit balance will drain with each request.
+
+If you want to continue using OCP without TUI-mode after 2026-06-15, budget for the Agent SDK credit cost accordingly — or switch to [OLP](https://github.com/dtzp555-max/olp) for multi-provider fallback.
+
+---
+
+## Verify after any flip
+
+1. Check `/health` shows the expected `tuiMode` state.
+2. Run the [615-canary](./615-canary.md) to confirm billing pool routing.
+3. If TUI-mode is ON: check `ocp logs 10` for any TUI spawn errors (`tui_spawn_failed`, tmux errors).
+
+---
+
+## Related
+
+- [615-canary runbook](./615-canary.md) — how to verify billing pool routing after a flip
+- [ADR 0007](../adr/0007-tui-interactive-mode.md) — TUI-mode architecture; Kill-switch section
+- [Subscription-pool (TUI) mode](../tui-mode.md#subscription-pool-tui-mode)
+- README § [Environment Variables](../../README.md#environment-variables) — `CLAUDE_TUI_MODE`, `OCP_TUI_ALLOW_LAN=1`

docs/superpowers/plans/2026-05-30-tui-mode-ocp-implementation-plan.md

@@ -0,0 +1,737 @@
+# TUI-mode (OCP-first) Implementation Plan
+
+> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.
+
+**Goal:** Add an opt-in `CLAUDE_TUI_MODE` to OCP that serves `/v1/chat/completions` by driving a *real interactive* `claude` session (no `-p`, no `--output-format`) so the request bills as `cc_entrypoint=cli` (subscription pool), reading the answer from claude's native JSONL transcript — while the default stream-json path stays byte-for-byte unchanged.
+
+**Architecture:** Two new pure-ish modules under `lib/tui/` — a transcript **reader** (`transcript.mjs`, provider-agnostic, the shareable core) and a tmux **session driver** (`session.mjs`, OCP-specific). `server.mjs` gains a `callClaudeTui()` that returns `Promise<string>` and is gated into the existing dispatch by a single env flag; because OCP's entire downstream (singleflight → `setCachedResponse` → `completionResponse` / chunked-SSE-replay → `recordUsage`) already consumes a string from `callClaude`, TUI-mode is a drop-in. Streaming is buffered then replayed as chunked SSE (no token streaming — deliberately, "don't build fragile features").
+
+**Tech Stack:** Node.js ESM (`.mjs`), `tmux` (interactive PTY host), `child_process` (`spawnSync`), `node:fs` polling (no `fs.watch`, no terminal-screen parsing). Test harness: `node test-features.mjs`.
+
+**Source of truth for the TUI mechanism:** the OLP design spec `docs/superpowers/specs/2026-05-30-tui-mode-production-design.md` (CLI-level, applies to both projects) + its 6 validation spikes (S1–S6, T1–T6) run on PI231 against `claude v2.1.158`. This plan is the OCP-grounded execution of that spec.
+
+---
+
+## Why OCP-first / scope decisions (read before coding)
+
+- **OCP-first** because OCP has the users and its compute path is `callClaude → Promise<string>`, a near-perfect impedance match for a reader that also returns a string. OLP would additionally need a string→IR-chunk-array adapter. OLP-sync is **deferred entirely until the post-2026-06-15 fork decision** — do not spend cycles keeping OLP's TUI in lockstep.
+- **A-path only.** Single-user / multi-device on one subscription. No per-key ephemeral isolation, no multi-tenant. (That is the OLP B-path, deferred.)
+- **A-path isolation = real `$HOME` + dedicated scratch cwd + `--strict-mcp-config`.** OCP has *no* ISOLATION contract and we do not build one. We run interactive `claude` in the operator's real home (OAuth + onboarding already valid) but in a **dedicated scratch working directory** (`OCP_TUI_CWD`, default `$HOME/.ocp-tui/work`) so transcripts land under one stable `projects/<cwd>` folder instead of polluting the operator's genuine project histories, and the trust-folder dialog is granted once.
+- **One `claude` session per request.** OCP is stateless (full conversation re-serialized each request via `messagesToPrompt`). TUI-mode mirrors this: per request, start a fresh interactive session with a fresh `--session-id`, submit one serialized prompt, await turn completion, read the transcript, extract the latest assistant text, tear the session down. Warm-pool / large-paste optimizations are explicitly out of v1 scope.
+- **Billing is unmeasurable until 2026-06-15.** Spike S1 proved the `cc_entrypoint=cli` *signal*, not the billed pool. The pre-6/15 deliverable is "a tested, working transport that emits `cli`"; 6/16 we flip the flag and measure with a documented kill-switch.
+- **Coexistence rule (PI231 runs an OLP test instance too).** All tmux sessions use the prefix `ocp-tui-`; the reaper kills **only** `ocp-tui-*`, never `olp-tui-*`. Never run two TUI proxies on the same OAuth concurrently — stop the OLP test instance during OCP integration.
+- **Provenance.** TUI-mode originated in OCP PR #101 (author courtesy: jaekwon-park <insainty21@gmail.com>). The PR #101 author should be credited + notified on the shipping PR.
+
+---
+
+## File Structure
+
+| File | Responsibility | New/Modified |
+|------|----------------|--------------|
+| `lib/tui/transcript.mjs` | Pure transcript parsing + the polling reader. Returns the latest assistant text once the turn is terminal or the wall-clock cap elapses. Provider-agnostic — the shareable core. | **Create** |
+| `lib/tui/session.mjs` | tmux session lifecycle: boot interactive `claude`, answer the trust dialog, submit the prompt (file → `"$(cat)"` paste → separate Enter), await the reader, tear down. Plus the prefix-scoped reaper. OCP-specific. | **Create** |
+| `lib/tui/fixtures/` | Real transcript JSONL harvested from PI231 + a few hand-crafted edge cases, for the reader's unit tests. | **Create** |
+| `server.mjs` | `callClaudeTui()` (`Promise<string>`); `streamStringAsSSE()` helper (DRY refactor of the cache-replay block); single-flag dispatch gates; reaper hook at boot; env consts. | **Modify** (`:258` env consts, `:1018`–`:1023` helpers, `:1467` dispatch, boot block) |
+| `test-features.mjs` | Suite for the reader (fixtures, runs in CI) + a live-only guarded suite for the driver (`OCP_TUI_LIVE=1`, skipped in CI). | **Modify** |
+| `docs/adr/0007-tui-interactive-mode.md` | OCP ADR 0007 (OCP's next number) — TUI mode rationale, billing-signal authority, scope, kill-switch. | **Create** |
+| `README.md` | New env vars (`CLAUDE_TUI_MODE`, `CLAUDE_TUI_WALLCLOCK_MS`, `OCP_TUI_CWD`), a "Subscription-pool (TUI) mode" section, troubleshooting + kill-switch. | **Modify** |
+| `CHANGELOG.md` | Unreleased entry. | **Modify** |
+
+---
+
+## PR-1 — Transcript reader (`lib/tui/transcript.mjs`)
+
+The shareable core. Pure functions + a polling reader. Fully unit-testable from committed fixtures; needs PI231 only once, to harvest realistic fixtures.
+
+### Task 0: Harvest real fixtures from PI231
+
+**Files:**
+- Create: `lib/tui/fixtures/complete-haiku.jsonl` (real, has `turn_duration`)
+- Create: `lib/tui/fixtures/complete-sonnet-multiblock.jsonl` (real, multi content-block answer)
+
+- [ ] **Step 1: Drive one real interactive turn on PI231 and copy its transcript**
+
+On PI231 (the only box with an authenticated interactive `claude`), run a single interactive turn in a scratch cwd, then locate its transcript:
+
+Run (on PI231):
+```bash
+SID=$(uuidgen)
+mkdir -p ~/.ocp-tui/work
+# drive one turn by hand in tmux OR reuse a transcript already produced by the S-spikes:
+ls -t ~/.claude/projects/-home-*-.ocp-tui-work/*.jsonl 2>/dev/null | head
+# pick one complete transcript (must contain a line with "subtype":"turn_duration")
+```
+Expected: at least one `.jsonl` file whose tail contains `{"type":"system","subtype":"turn_duration",...}`.
+
+- [ ] **Step 2: Copy 2 real transcripts into the repo as fixtures, scrubbed**
+
+Run (from the workstation):
+```bash
+scp pi231:'~/.claude/projects/<encoded-cwd>/<sid>.jsonl' lib/tui/fixtures/complete-haiku.jsonl
+# Scrub: the transcript may contain the prompt/answer text only (no OAuth token — tokens
+# live in ~/.claude/.credentials.json, NOT in projects/*.jsonl). Confirm no credential
+# material before committing:
+grep -iE "sk-ant|oat01|bearer|authorization" lib/tui/fixtures/*.jsonl && echo "STOP: scrub" || echo "clean"
+```
+Expected: `clean`. (Transcripts hold conversation content + metadata, never the bearer token. If a fixture's prompt text is sensitive, replace it with a benign hand-edited turn that keeps the JSON shape.)
+
+- [ ] **Step 3: Commit the fixtures**
+
+```bash
+git add lib/tui/fixtures/complete-haiku.jsonl lib/tui/fixtures/complete-sonnet-multiblock.jsonl
+git commit -m "test(tui): real claude transcript fixtures harvested from PI231 (v2.1.158)"
+```
+
+### Task 1: `encodeCwd` + `transcriptPath` (the path formula)
+
+**Files:**
+- Create: `lib/tui/transcript.mjs`
+- Test: `test-features.mjs` (new Suite "TUI transcript")
+
+- [ ] **Step 1: Write the failing test**
+
+Add to `test-features.mjs`:
+```js
+// ── Suite: TUI transcript reader ────────────────────────────────────────
+import { encodeCwd, transcriptPath } from "./lib/tui/transcript.mjs";
+
+test("encodeCwd replaces every slash incl. leading", () => {
+  assertEqual(encodeCwd("/home/u/.ocp-tui/work"), "-home-u-.ocp-tui-work");
+});
+test("transcriptPath composes EHOME/.claude/projects/<enc>/<sid>.jsonl", () => {
+  assertEqual(
+    transcriptPath("/home/u", "/home/u/.ocp-tui/work", "abc-123"),
+    "/home/u/.claude/projects/-home-u-.ocp-tui-work/abc-123.jsonl"
+  );
+});
+```
+
+- [ ] **Step 2: Run to verify it fails**
+
+Run: `node test-features.mjs 2>&1 | grep -i "tui transcript\|Cannot find module"`
+Expected: FAIL — `Cannot find module './lib/tui/transcript.mjs'`.
+
+- [ ] **Step 3: Minimal implementation**
+
+Create `lib/tui/transcript.mjs`:
+```js
+// Transcript reader for TUI-mode. Reads claude's native JSONL session transcript
+// and returns the latest assistant turn's text once the turn is terminal.
+//
+// Authority: claude CLI v2.1.158 — interactive session transcript at
+//   <HOME>/.claude/projects/<CWD with every "/" -> "-">/<--session-id>.jsonl
+// Completion marker: a line {"type":"system","subtype":"turn_duration",...}.
+// See docs/superpowers/specs/2026-05-30-tui-mode-production-design.md §4.
+import { readFileSync, existsSync } from "node:fs";
+
+const sleep = (ms) => new Promise((r) => setTimeout(r, ms));
+
+// Project-dir encoding: every "/" -> "-" (including the leading slash).
+export function encodeCwd(cwd) {
+  return cwd.replace(/\//g, "-");
+}
+
+export function transcriptPath(home, cwd, sessionId) {
+  return `${home}/.claude/projects/${encodeCwd(cwd)}/${sessionId}.jsonl`;
+}
+```
+
+- [ ] **Step 4: Run to verify it passes**
+
+Run: `node test-features.mjs 2>&1 | grep -i "tui transcript"`
+Expected: PASS for both cases.
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add lib/tui/transcript.mjs test-features.mjs
+git commit -m "feat(tui): transcript path formula (encodeCwd + transcriptPath)"
+```
+
+### Task 2: `parseTranscriptLines` + `isTerminalLine` + `extractLatestAssistantText`
+
+**Files:**
+- Modify: `lib/tui/transcript.mjs`
+- Test: `test-features.mjs`
+
+- [ ] **Step 1: Write the failing tests**
+
+```js
+import { parseTranscriptLines, isTerminalLine, extractLatestAssistantText } from "./lib/tui/transcript.mjs";
+import { readFileSync } from "node:fs";
+
+test("parseTranscriptLines skips blank + malformed/partial lines", () => {
+  const evs = parseTranscriptLines('{"a":1}\n\n{bad json\n{"b":2}\n');
+  assertEqual(evs.length, 2);
+  assertEqual(evs[1].b, 2);
+});
+test("isTerminalLine true on turn_duration", () => {
+  assertEqual(isTerminalLine({ type: "system", subtype: "turn_duration" }), true);
+});
+test("isTerminalLine true on stop_reason tool_use (message-wrapped + flat)", () => {
+  assertEqual(isTerminalLine({ type: "assistant", message: { stop_reason: "tool_use" } }), true);
+  assertEqual(isTerminalLine({ stop_reason: "tool_use" }), true);
+});
+test("isTerminalLine false on ordinary assistant/text lines", () => {
+  assertEqual(isTerminalLine({ type: "assistant", message: { content: [{ type: "text", text: "hi" }] } }), false);
+});
+test("extractLatestAssistantText concatenates text blocks of the LAST assistant turn", () => {
+  const evs = [
+    { type: "assistant", message: { content: [{ type: "text", text: "first" }] } },
+    { type: "user", message: { content: "..." } },
+    { type: "assistant", message: { content: [{ type: "text", text: "A" }, { type: "thinking", thinking: "x" }, { type: "text", text: "B" }] } },
+  ];
+  assertEqual(extractLatestAssistantText(evs), "AB");
+});
+test("real complete fixture yields non-empty text and is terminal", () => {
+  const evs = parseTranscriptLines(readFileSync("./lib/tui/fixtures/complete-haiku.jsonl", "utf8"));
+  assert(evs.some(isTerminalLine), "fixture must contain a terminal line");
+  assert(extractLatestAssistantText(evs).length > 0, "fixture must yield assistant text");
+});
+```
+
+- [ ] **Step 2: Run to verify it fails**
+
+Run: `node test-features.mjs 2>&1 | grep -i "parseTranscript\|isTerminal\|extractLatest\|real complete fixture"`
+Expected: FAIL — exports not defined.
+
+- [ ] **Step 3: Minimal implementation** (append to `lib/tui/transcript.mjs`)
+
+```js
+// Parse NDJSON text into objects; skip blank lines and partial/forming lines
+// (the live transcript is read mid-write, so the last line may be incomplete).
+export function parseTranscriptLines(text) {
+  const out = [];
+  for (const line of text.split("\n")) {
+    const t = line.trim();
+    if (!t) continue;
+    try { out.push(JSON.parse(t)); } catch { /* partial line being written */ }
+  }
+  return out;
+}
+
+// A line marks the assistant turn complete when it is the turn_duration system
+// event, or an assistant message that stopped to hand off to a tool.
+export function isTerminalLine(obj) {
+  if (!obj || typeof obj !== "object") return false;
+  if (obj.type === "system" && obj.subtype === "turn_duration") return true;
+  const sr = (obj.message && obj.message.stop_reason) || obj.stop_reason;
+  return sr === "tool_use";
+}
+
+// Text of the LAST assistant turn: concatenate its text content blocks
+// (ignore thinking/tool_use blocks). Later assistant entries overwrite earlier.
+export function extractLatestAssistantText(events) {
+  let text = "";
+  for (const ev of events) {
+    if (!ev || ev.type !== "assistant") continue;
+    const content = ev.message && ev.message.content;
+    if (!Array.isArray(content)) continue;
+    const parts = content
+      .filter((b) => b && b.type === "text" && typeof b.text === "string")
+      .map((b) => b.text);
+    if (parts.length) text = parts.join("");
+  }
+  return text;
+}
+```
+
+- [ ] **Step 4: Run to verify it passes**
+
+Run: `node test-features.mjs 2>&1 | grep -iE "parseTranscript|isTerminal|extractLatest|real complete fixture"`
+Expected: all PASS.
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add lib/tui/transcript.mjs test-features.mjs
+git commit -m "feat(tui): transcript parsing + terminal detection + assistant-text extraction"
+```
+
+### Task 3: `readTuiTranscript` (the polling reader with wall-clock cap)
+
+**Files:**
+- Modify: `lib/tui/transcript.mjs`
+- Test: `test-features.mjs`
+
+- [ ] **Step 1: Write the failing tests**
+
+```js
+import { readTuiTranscript } from "./lib/tui/transcript.mjs";
+import { mkdtempSync, writeFileSync } from "node:fs";
+import { tmpdir } from "node:os";
+
+test("readTuiTranscript returns assistant text when terminal marker present", async () => {
+  const dir = mkdtempSync(`${tmpdir()}/tui-`);
+  const p = `${dir}/s.jsonl`;
+  writeFileSync(p, [
+    JSON.stringify({ type: "assistant", message: { content: [{ type: "text", text: "hello world" }] } }),
+    JSON.stringify({ type: "system", subtype: "turn_duration", durationMs: 1200 }),
+  ].join("\n") + "\n");
+  const out = await readTuiTranscript({ transcriptPath: p, wallclockMs: 2000, pollMs: 50 });
+  assertEqual(out, "hello world");
+});
+test("readTuiTranscript honours wall-clock cap and returns partial text", async () => {
+  const dir = mkdtempSync(`${tmpdir()}/tui-`);
+  const p = `${dir}/s.jsonl`;
+  writeFileSync(p, JSON.stringify({ type: "assistant", message: { content: [{ type: "text", text: "partial" }] } }) + "\n");
+  const out = await readTuiTranscript({ transcriptPath: p, wallclockMs: 300, pollMs: 50 }); // never terminal
+  assertEqual(out, "partial");
+});
+test("readTuiTranscript throws when no text and cap elapses", async () => {
+  const dir = mkdtempSync(`${tmpdir()}/tui-`);
+  const p = `${dir}/missing.jsonl`; // file never appears
+  let threw = false;
+  try { await readTuiTranscript({ transcriptPath: p, wallclockMs: 200, pollMs: 50 }); }
+  catch { threw = true; }
+  assert(threw, "must throw on empty timeout");
+});
+```
+
+- [ ] **Step 2: Run to verify it fails**
+
+Run: `node test-features.mjs 2>&1 | grep -i "readTuiTranscript"`
+Expected: FAIL — export not defined.
+
+- [ ] **Step 3: Minimal implementation** (append)
+
+```js
+// Block until the session transcript is terminal (turn_duration / tool_use) or
+// the wall-clock cap elapses, polling the file (no fs.watch — robust over NFS /
+// editors). Returns the latest assistant text. On cap with text, returns the
+// partial text; on cap with no text at all, throws.
+//
+// No quiescence heuristic by design: a long Opus thinking turn stalls transcript
+// growth and a "file stable for N s" rule would false-abort it (spec §4.3).
+export async function readTuiTranscript({ transcriptPath: p, wallclockMs = 120000, pollMs = 250 }) {
+  const deadline = Date.now() + wallclockMs;
+  let lastText = "";
+  while (Date.now() < deadline) {
+    if (existsSync(p)) {
+      const events = parseTranscriptLines(readFileSync(p, "utf8"));
+      lastText = extractLatestAssistantText(events) || lastText;
+      if (events.some(isTerminalLine)) return lastText;
+    }
+    await sleep(pollMs);
+  }
+  if (lastText) return lastText;
+  throw new Error("tui_transcript_timeout: no assistant text within wallclock cap");
+}
+```
+
+- [ ] **Step 4: Run to verify it passes**
+
+Run: `node test-features.mjs 2>&1 | grep -i "readTuiTranscript"`
+Expected: all 3 PASS.
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add lib/tui/transcript.mjs test-features.mjs
+git commit -m "feat(tui): polling transcript reader with wall-clock cap (no quiescence)"
+```
+
+---
+
+## PR-2 — Session driver (`lib/tui/session.mjs`)
+
+tmux lifecycle + the validated submission recipe. Cannot be unit-tested without a live authenticated `claude`; tested by a live-only guarded suite that runs on PI231.
+
+### Task 4: `reapStaleTuiSessions` (prefix-scoped reaper)
+
+**Files:**
+- Create: `lib/tui/session.mjs`
+- Test: `test-features.mjs`
+
+- [ ] **Step 1: Write the failing test** (pure — no live claude; inject a fake tmux runner)
+
+```js
+import { reapStaleTuiSessions, SESSION_PREFIX } from "./lib/tui/session.mjs";
+
+test("reaper kills ONLY ocp-tui- sessions, never olp-tui-", () => {
+  const killed = [];
+  const fakeTmux = (args) => {
+    if (args[0] === "list-sessions") return { status: 0, stdout: "ocp-tui-aaaa\nolp-tui-bbbb\nmisc\nocp-tui-cccc\n" };
+    if (args[0] === "kill-session") { killed.push(args[args.indexOf("-t") + 1]); return { status: 0 }; }
+    return { status: 0, stdout: "" };
+  };
+  const n = reapStaleTuiSessions({ tmux: fakeTmux });
+  assertEqual(SESSION_PREFIX, "ocp-tui-");
+  assertEqual(n, 2);
+  assertEqual(killed.join(","), "ocp-tui-aaaa,ocp-tui-cccc");
+});
+```
+
+- [ ] **Step 2: Run to verify it fails**
+
+Run: `node test-features.mjs 2>&1 | grep -i "reaper kills"`
+Expected: FAIL — module/export missing.
+
+- [ ] **Step 3: Minimal implementation**
+
+Create `lib/tui/session.mjs`:
+```js
+// TUI-mode session driver: hosts an interactive `claude` in tmux, submits one
+// serialized prompt, awaits the transcript reader, tears down. OCP-specific.
+//
+// Authority: claude CLI v2.1.158 interactive mode (no -p / no --output-format
+// => cc_entrypoint=cli). Submission recipe + dialog handling validated by spikes
+// T3/T6 on PI231. See docs/superpowers/specs/2026-05-30-tui-mode-production-design.md.
+import { spawnSync } from "node:child_process";
+import { mkdtempSync, writeFileSync, mkdirSync, existsSync, rmSync } from "node:fs";
+import { tmpdir } from "node:os";
+import { randomUUID } from "node:crypto";
+import { transcriptPath, readTuiTranscript } from "./transcript.mjs";
+
+export const SESSION_PREFIX = "ocp-tui-"; // per-proxy namespace (coexistence rule)
+const TMUX = process.env.OCP_TUI_TMUX_BIN || "tmux";
+
+const sleep = (ms) => new Promise((r) => setTimeout(r, ms));
+const defaultTmux = (args, opts = {}) => spawnSync(TMUX, args, { encoding: "utf8", ...opts });
+
+// Kill ONLY our own stale sessions. Scoped to SESSION_PREFIX so a co-hosted
+// OLP test instance's `olp-tui-*` sessions are never touched.
+export function reapStaleTuiSessions({ tmux = defaultTmux } = {}) {
+  const r = tmux(["list-sessions", "-F", "#{session_name}"]);
+  if (!r || r.status !== 0) return 0; // no tmux server / no sessions
+  let killed = 0;
+  for (const name of String(r.stdout || "").split("\n").map((s) => s.trim()).filter(Boolean)) {
+    if (name.startsWith(SESSION_PREFIX)) { tmux(["kill-session", "-t", name]); killed++; }
+  }
+  return killed;
+}
+```
+
+- [ ] **Step 4: Run to verify it passes**
+
+Run: `node test-features.mjs 2>&1 | grep -i "reaper kills"`
+Expected: PASS.
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add lib/tui/session.mjs test-features.mjs
+git commit -m "feat(tui): prefix-scoped session reaper (ocp-tui-* only)"
+```
+
+### Task 5: `runTuiTurn` (boot → trust dialog → paste → Enter → read → teardown)
+
+**Files:**
+- Modify: `lib/tui/session.mjs`
+- Test: `test-features.mjs` (live-only, guarded by `OCP_TUI_LIVE=1`)
+
+- [ ] **Step 1: Write the live-only guarded test** (skipped in CI; run on PI231)
+
+```js
+// Live-only: requires an authenticated interactive `claude`. Skipped unless OCP_TUI_LIVE=1.
+if (process.env.OCP_TUI_LIVE === "1") {
+  test("runTuiTurn drives a real interactive turn and returns text", async () => {
+    const { runTuiTurn } = await import("./lib/tui/session.mjs");
+    const out = await runTuiTurn({
+      prompt: "Reply with exactly the word PONG and nothing else.",
+      model: "claude-haiku-4-5-20251001",
+      claudeBin: process.env.OCP_TUI_CLAUDE_BIN || "claude",
+      home: process.env.HOME,
+      cwd: `${process.env.HOME}/.ocp-tui/work`,
+      wallclockMs: 120000,
+    });
+    assert(/PONG/i.test(out), `expected PONG, got: ${out.slice(0, 200)}`);
+  });
+} else {
+  test("runTuiTurn (live) — SKIPPED (set OCP_TUI_LIVE=1 on PI231 to run)", () => { assert(true); });
+}
+```
+
+- [ ] **Step 2: Run to verify it fails** (on a box, with the flag)
+
+Run (PI231): `OCP_TUI_LIVE=1 node test-features.mjs 2>&1 | grep -i "runTuiTurn"`
+Expected: FAIL — `runTuiTurn` not exported yet.
+
+- [ ] **Step 3: Implementation** (append to `lib/tui/session.mjs`)
+
+```js
+// Boot wait + dialog timing. Conservative defaults validated on PI231; env-tunable.
+const BOOT_MS   = parseInt(process.env.OCP_TUI_BOOT_MS   || "3500", 10);
+const DIALOG_MS = parseInt(process.env.OCP_TUI_DIALOG_MS || "1200", 10);
+const PASTE_SETTLE_MS = parseInt(process.env.OCP_TUI_PASTE_MS || "1800", 10);
+
+const shq = (s) => `'${String(s).replace(/'/g, "'\\''")}'`; // single-quote for sh -c
+
+// Build interactive claude argv: NO -p, NO --output-format (=> cc_entrypoint=cli).
+// MCP hard-disabled: --strict-mcp-config (no --mcp-config) is the only mechanism
+// that stops account-attached managed MCP from connecting (spec §5.2 / T6),
+// belt-and-braces with --disallowedTools "mcp__*".
+function buildTuiCmd(claudeBin, model, sessionId) {
+  return [
+    shq(claudeBin),
+    "--model", shq(model),
+    "--session-id", sessionId,
+    "--strict-mcp-config",
+    "--disallowedTools", shq("mcp__*"),
+  ].join(" ");
+}
+
+export async function runTuiTurn({
+  prompt, model, claudeBin, home, cwd,
+  wallclockMs = 120000, tmux = defaultTmux,
+}) {
+  const sessionId = randomUUID();
+  const tmuxName = SESSION_PREFIX + sessionId.slice(0, 8);
+  if (!existsSync(cwd)) mkdirSync(cwd, { recursive: true });
+
+  const tmpDir = mkdtempSync(`${tmpdir()}/ocp-tui-`);
+  const promptFile = `${tmpDir}/prompt.txt`;
+  writeFileSync(promptFile, prompt, { mode: 0o600 });
+
+  const env = { ...process.env, CLAUDE_CODE_DISABLE_OFFICIAL_MARKETPLACE_AUTOINSTALL: "1" };
+  delete env.CLAUDECODE; delete env.ANTHROPIC_API_KEY; delete env.ANTHROPIC_BASE_URL; delete env.ANTHROPIC_AUTH_TOKEN;
+  if (home) env.HOME = home;
+
+  try {
+    // 1. Boot the interactive session inside tmux, in the dedicated scratch cwd.
+    tmux(["new-session", "-d", "-s", tmuxName, "-x", "220", "-y", "50", "-c", cwd,
+          buildTuiCmd(claudeBin, model, sessionId)], { env });
+    await sleep(BOOT_MS);
+
+    // 2. Answer the trust-folder dialog defensively. The seeded bypass flag (if any)
+    //    suppresses the *bypass-permissions* dialog but NOT the trust-folder dialog;
+    //    "1" = "Yes, proceed". Harmless if the dialog is absent (cwd already trusted).
+    tmux(["send-keys", "-t", tmuxName, "1"]);
+    tmux(["send-keys", "-t", tmuxName, "Enter"]);
+    await sleep(DIALOG_MS);
+
+    // 3. Submit the prompt. Body is pasted via `"$(cat file)"` so the content never
+    //    touches the command line (no shell injection from prompt text), then a
+    //    SEPARATE Enter key event submits it (Ink #15553: literal "\n" in a paste
+    //    does not submit; the Enter key event does).
+    spawnSync("sh", ["-c",
+      `${shq(TMUX)} send-keys -t ${shq(tmuxName)} -- "$(cat ${shq(promptFile)})"`],
+      { env, encoding: "utf8" });
+    await sleep(PASTE_SETTLE_MS);
+    tmux(["send-keys", "-t", tmuxName, "Enter"]);
+
+    // 4. Read the answer from the native transcript.
+    const tpath = transcriptPath(home || process.env.HOME, cwd, sessionId);
+    return await readTuiTranscript({ transcriptPath: tpath, wallclockMs });
+  } finally {
+    // 5. Teardown — always. Kill the session, remove the temp prompt dir.
+    try { tmux(["kill-session", "-t", tmuxName]); } catch { /* already gone */ }
+    try { rmSync(tmpDir, { recursive: true, force: true }); } catch { /* best effort */ }
+  }
+}
+```
+
+- [ ] **Step 4: Run to verify it passes** (PI231, live)
+
+Run (PI231): `OCP_TUI_LIVE=1 node test-features.mjs 2>&1 | grep -i "runTuiTurn"`
+Expected: PASS — output contains `PONG`. Also confirm no orphan sessions: `tmux ls 2>/dev/null | grep ocp-tui- || echo "clean"` → `clean`.
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add lib/tui/session.mjs test-features.mjs
+git commit -m "feat(tui): runTuiTurn — interactive session driver (boot/trust/paste/Enter/read/teardown)"
+```
+
+---
+
+## PR-3 — Wiring into `server.mjs`
+
+Gate TUI-mode behind one env flag. Default path (`CLAUDE_TUI_MODE` unset) stays byte-for-byte identical.
+
+### Task 6: env consts + `streamStringAsSSE` DRY refactor
+
+**Files:**
+- Modify: `server.mjs` (env consts near `:275`; refactor cache-replay block `:1524`–`:1539` into a helper near `:1023`)
+
+- [ ] **Step 1: Add TUI env consts + import** (near the other `const ... = process.env...` at `server.mjs:258`–`:275`)
+
+```js
+import { runTuiTurn, reapStaleTuiSessions } from "./lib/tui/session.mjs";
+
+// TUI-mode (subscription-pool bridge). Opt-in; default OFF keeps stream-json path.
+// Authority: docs/adr/0007-tui-interactive-mode.md.
+const TUI_MODE = process.env.CLAUDE_TUI_MODE === "true";
+const TUI_WALLCLOCK_MS = parseInt(process.env.CLAUDE_TUI_WALLCLOCK_MS || "120000", 10);
+const TUI_CWD = process.env.OCP_TUI_CWD || `${process.env.HOME}/.ocp-tui/work`;
+```
+
+- [ ] **Step 2: Extract the chunked-SSE-replay into a reusable helper** (near `completionResponse` at `:1023`)
+
+```js
+// Replay a complete string as a chunked SSE stream (80 codepoints/chunk).
+// Extracted from the cache-hit replay block so TUI-mode streaming reuses it.
+function streamStringAsSSE(res, id, model, content) {
+  const created = Math.floor(Date.now() / 1000);
+  res.writeHead(200, { "Content-Type": "text/event-stream", "Cache-Control": "no-cache", "Connection": "keep-alive", "X-Accel-Buffering": "no" });
+  sendSSE(res, { id, object: "chat.completion.chunk", created, model, choices: [{ index: 0, delta: { role: "assistant" }, finish_reason: null }] });
+  const CHUNK = 80;
+  const codepoints = Array.from(content);
+  for (let i = 0; i < codepoints.length; i += CHUNK) {
+    sendSSE(res, { id, object: "chat.completion.chunk", created, model, choices: [{ index: 0, delta: { content: codepoints.slice(i, i + CHUNK).join("") }, finish_reason: null }] });
+  }
+  sendSSE(res, { id, object: "chat.completion.chunk", created, model, choices: [{ index: 0, delta: {}, finish_reason: "stop" }] });
+  res.write("data: [DONE]\n\n");
+  res.end();
+}
+```
+
+- [ ] **Step 3: Point the cache-hit streaming replay (`:1524`–`:1539`) at the helper** (DRY — behavior identical)
+
+Replace the inline block inside `if (stream) { ... }` of the cache hit with:
+```js
+          if (stream) {
+            const id = `chatcmpl-${randomUUID()}`;
+            streamStringAsSSE(res, id, model, cached.response);
+            return;
+          } else {
+```
+
+- [ ] **Step 4: Run the full suite to verify no regression**
+
+Run: `node test-features.mjs 2>&1 | tail -3`
+Expected: all existing tests PASS (the refactor is behavior-preserving; cache-replay covered by existing D3 tests).
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add server.mjs
+git commit -m "refactor(server): extract streamStringAsSSE helper + add TUI env consts"
+```
+
+### Task 7: `callClaudeTui` + dispatch gates
+
+**Files:**
+- Modify: `server.mjs` (new `callClaudeTui` near `callClaude:735`; gates at the buffered dispatch `:1563`/`:1594` and streaming dispatch `:1551`)
+
+- [ ] **Step 1: Add `callClaudeTui`** (near `callClaude`, after `:800`)
+
+```js
+// TUI-mode upstream: drive an interactive claude session, return the assistant
+// text as a string — same contract as callClaude(), so all downstream
+// (singleflight, cache write-back, completionResponse) is unchanged.
+// System messages are rendered inline as [System] blocks by messagesToPrompt;
+// we deliberately do NOT pass --system-prompt in interactive mode to avoid any
+// flag that could perturb cc_entrypoint classification.
+function callClaudeTui(model, messages, conversationId, keyName) {
+  const cliModel = MODEL_MAP[model] || model;
+  const prompt = messagesToPrompt(messages); // includes system as [System] inline
+  recordModelRequest(cliModel, prompt.length);
+  return runTuiTurn({
+    prompt, model: cliModel, claudeBin: CLAUDE,
+    home: process.env.HOME, cwd: TUI_CWD, wallclockMs: TUI_WALLCLOCK_MS,
+  }).then((text) => {
+    recordModelSuccess(cliModel, 0);
+    return text;
+  }).catch((err) => {
+    recordModelError(cliModel, false);
+    throw err;
+  });
+}
+```
+
+- [ ] **Step 2: Gate the buffered dispatch** — at `server.mjs:1563`–`:1597`, replace the two `callClaude(...)` call sites (inside the singleflight closure and the cache-disabled fallback) with a selected upstream:
+
+Add once, just before the `if (CACHE_TTL > 0 && req._cacheHash)` block (~`:1563`):
+```js
+  const upstreamCall = TUI_MODE ? callClaudeTui : callClaude;
+```
+Then change `await callClaude(model, messages, conversationId, req._authKeyName)` → `await upstreamCall(model, messages, conversationId, req._authKeyName)` at **both** sites (`:1572` and `:1594`).
+
+- [ ] **Step 3: Gate the streaming dispatch** — at `server.mjs:1551`–`:1553`, branch TUI streaming to buffer-then-replay:
+
+```js
+  if (stream) {
+    if (TUI_MODE) {
+      // TUI has no token stream; buffer the turn, write-back to cache, replay as chunked SSE.
+      const t0Usage = Date.now();
+      try {
+        const content = await callClaudeTui(model, messages, conversationId, req._authKeyName);
+        if (CACHE_TTL > 0 && req._cacheHash) {
+          try { setCachedResponse(req._cacheHash, model, content); } catch (e) { logEvent("error", "cache_write_failed", { error: e.message }); }
+        }
+        const id = `chatcmpl-${randomUUID()}`;
+        streamStringAsSSE(res, id, model, content);
+        try { recordUsage({ keyId: req._authKeyId, keyName: req._authKeyName, model, promptChars: messages.reduce((a, m) => a + (typeof m.content === "string" ? m.content.length : JSON.stringify(m.content).length), 0), responseChars: content.length, elapsedMs: Date.now() - t0Usage, success: true }); } catch {}
+        return;
+      } catch (err) {
+        if (res.headersSent || res.writableEnded || res.destroyed) { try { res.end(); } catch {}; return; }
+        const safeMessage = (err.message || "Internal error").replace(/\/[\w/.\-]+/g, "[path]");
+        return jsonResponse(res, 500, { error: { message: safeMessage, type: "proxy_error" } });
+      }
+    }
+    // Default: real stream-json streaming, unchanged.
+    return callClaudeStreaming(model, messages, conversationId, res, { keyId: req._authKeyId, keyName: req._authKeyName, cacheHash: req._cacheHash });
+  }
+```
+
+- [ ] **Step 4: Verify default path is untouched + TUI path selected only by flag**
+
+Run: `CLAUDE_TUI_MODE=  node -e "process.env.CLAUDE_TUI_MODE; import('./server.mjs')" 2>&1 | head -1 || true`
+Then the regression suite: `node test-features.mjs 2>&1 | tail -3`
+Expected: all PASS (no test sets `CLAUDE_TUI_MODE`, so `upstreamCall === callClaude` and streaming uses `callClaudeStreaming` — identical to today).
+
+Live end-to-end (PI231, after Task 8 setup): with `CLAUDE_TUI_MODE=true` start OCP and `curl` both `stream:false` and `stream:true`:
+```bash
+curl -s localhost:3456/v1/chat/completions -H "Authorization: Bearer <key>" \
+  -d '{"model":"claude-haiku-4-5-20251001","messages":[{"role":"user","content":"say PONG"}]}' | head
+```
+Expected: a normal OpenAI completion whose content contains `PONG`. Cross-check on PI231 that the spawned `claude` had no `-p`/`--output-format` (`ps -ef | grep claude`).
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add server.mjs
+git commit -m "feat(tui): gate interactive TUI upstream behind CLAUDE_TUI_MODE (buffered + streaming)"
+```
+
+### Task 8: reaper hook at boot + ADR + README + CHANGELOG
+
+**Files:**
+- Modify: `server.mjs` (boot block — call `reapStaleTuiSessions()` once on startup when `TUI_MODE`)
+- Create: `docs/adr/0007-tui-interactive-mode.md`
+- Modify: `README.md`, `CHANGELOG.md`
+
+- [ ] **Step 1: Reaper on boot** (in the server start/`listen` block)
+
+```js
+  if (TUI_MODE) {
+    try { const n = reapStaleTuiSessions(); if (n) logEvent("info", "tui_reaped_stale_sessions", { count: n }); } catch {}
+    console.log(`  TUI-mode: ON (interactive claude → cc_entrypoint=cli). cwd=${TUI_CWD} wallclock=${TUI_WALLCLOCK_MS}ms`);
+  }
+```
+
+- [ ] **Step 2: Write ADR 0007** — `docs/adr/0007-tui-interactive-mode.md`
+
+Context: 2026-06-15 billing split routes by `cc_entrypoint`; `-p`/`--output-format` ⇒ `sdk-cli` (Agent SDK credit pool, ~$20 on Pro = unusable). Decision: opt-in interactive driver ⇒ `cli` (subscription pool). Authority: spec §1/§4, claude v2.1.158. Scope: A-path single-user; MCP hard-disabled via `--strict-mcp-config`. Kill-switch: unset `CLAUDE_TUI_MODE` → stream-json path restored. Consequences: no token streaming (buffered+replayed); grey-area, billing unmeasurable until 6/15; reaper + tmux-prefix coexistence rules.
+
+- [ ] **Step 3: README** — add `CLAUDE_TUI_MODE`, `CLAUDE_TUI_WALLCLOCK_MS`, `OCP_TUI_CWD` to the env-var table; add a "Subscription-pool (TUI) mode" section (what it is, opt-in, the 6/15 rationale, no-streaming caveat, the one-time `mkdir -p ~/.ocp-tui/work` + tmux dependency, and the `CLAUDE_TUI_MODE` unset kill-switch).
+
+- [ ] **Step 4: CHANGELOG** — Unreleased: `feat(tui): opt-in CLAUDE_TUI_MODE — serve via interactive claude (cc_entrypoint=cli / subscription pool); default stream-json path unchanged.`
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add server.mjs docs/adr/0007-tui-interactive-mode.md README.md CHANGELOG.md
+git commit -m "feat(tui): boot reaper + ADR 0007 + README + CHANGELOG (TUI-mode docs)"
+```
+
+---
+
+## Integration & canary (post-implementation, on PI231)
+
+1. Stop the OLP test instance (`:4567`) — clean shared OAuth + no tmux collision.
+2. `git clone`/checkout this branch on PI231, `mkdir -p ~/.ocp-tui/work`, start OCP on `:3456` with `CLAUDE_TUI_MODE=true`.
+3. Run the live driver suite: `OCP_TUI_LIVE=1 node test-features.mjs`.
+4. End-to-end `curl` (buffered + streaming) through OCP; confirm spawned `claude` carries no `-p`/`--output-format`.
+5. **Pre-6/15 deliverable = here.** Billing measurement waits for 6/15; document the kill-switch (unset `CLAUDE_TUI_MODE`).
+
+---
+
+## Self-Review (against spec + the OCP-first execution review)
+
+- **Spec coverage:** transcript path formula (§4 → Task 1), parsing/terminal/extract (§4 → Task 2), polling reader + wall-clock cap + no-quiescence (§4.3 → Task 3), submission recipe file→paste→Enter (§5/T3 → Task 5), trust-dialog handling (§5.2 → Task 5), MCP disable `--strict-mcp-config` (§5.2/T6 → Tasks 5 & buildTuiCmd), string-contract drop-in (→ Tasks 6–7), kill-switch + default-path-sacred (→ Task 7 Step 4), coexistence prefix + reaper (→ Tasks 4 & 8). ✅
+- **Review findings folded:** OCP-first string match (Task 7); no ephemeral-home, real-home + scratch cwd (scope §); reader-only sharing, driver forked (file table); tmux prefix + scoped reaper + never-both-on-OAuth (Task 4, Integration §1); `TIMEOUT=600000 > 120s` cap verified (no SIGKILL-mid-turn); `--strict-mcp-config` added (Task 5); provenance jaekwon-park (Why §). OLP-sync deferred. ✅
+- **Placeholder scan:** none — every code step carries real code; every run step an exact command + expected output. ✅
+- **Type consistency:** `runTuiTurn`/`reapStaleTuiSessions`/`SESSION_PREFIX` exported in Task 4–5 match imports in Task 6–8; `streamStringAsSSE(res, id, model, content)` defined Task 6, used Tasks 6–7; `callClaudeTui(model, messages, conversationId, keyName)` mirrors `callClaude`'s signature. ✅
+- **Open item for integration:** confirm on PI231 that the seeded `~/.claude.json` is unnecessary for real-home A (onboarding already complete); if a bypass-permissions dialog *does* appear in real home, add a one-line seed step (`bypassPermissionsModeAccepted:true`) — but the driver already answers the trust dialog defensively, so the turn still completes.

docs/troubleshooting.md

@@ -0,0 +1,136 @@
+Part of [OCP](../README.md) — full troubleshooting manual. The README keeps a slim version with the most common issues and the one-time bootstrap quirks; everything else lives here.
+
+# Troubleshooting
+
+The simplest path: ask your AI.
+
+  Paste this prompt:
+
+  ```
+  Run `ocp doctor` and follow its `next_action`. Tell me if you hit
+  anything that needs human input.
+  ```
+
+The doctor produces a JSON `next_action` with `ai_executable[]` (commands
+the agent runs verbatim) and `human_required[]` (steps that need you,
+typically just OAuth).
+
+## Manual debugging
+
+### Setup fails with "claude: command not found"
+
+`setup.mjs` requires the Claude CLI to be on `PATH`. Install it via the [official guide](https://docs.anthropic.com/en/docs/claude-cli), confirm with `which claude`, then run `claude auth login` before re-running `node setup.mjs`.
+
+### Setup fails with "EADDRINUSE: port 3456 already in use"
+
+Something else is already bound to port 3456 — usually an old OCP instance. Check what:
+
+```bash
+lsof -nP -iTCP:3456 -sTCP:LISTEN
+```
+
+If it's an old OCP process, stop it before re-running setup:
+
+```bash
+launchctl bootout gui/$(id -u)/dev.ocp.proxy            # macOS launchd
+systemctl --user stop ocp-proxy                         # Linux systemd (installed as a --user unit)
+```
+
+(There is no `ocp stop` subcommand — the proxy runs as a service, so stopping it goes through the service manager above. `ocp restart` exists for the bounce case.)
+
+### Setup fails with "node: command not found" or version error
+
+OCP requires Node.js 22.5+. Install:
+
+```bash
+brew install node          # macOS
+# Linux: see https://nodejs.org/en/download for current install commands
+```
+
+Confirm with `node --version` (should be ≥ v22.5).
+
+### Requests fail or agents stuck
+
+```bash
+# Clear sessions and restart
+ocp clear
+ocp restart
+
+# If using OpenClaw gateway
+openclaw gateway restart
+```
+
+### Env var change (e.g. `CLAUDE_BIND`, `CLAUDE_CODE_OAUTH_TOKEN`) doesn't take effect after restart
+
+On **macOS**, `ocp restart` does a full `launchctl bootout` + `bootstrap` of the agent, which **re-reads the plist `EnvironmentVariables`** — so an env change you made (in `~/Library/LaunchAgents/dev.ocp.proxy.plist`) actually takes effect:
+
+```bash
+ocp restart
+```
+
+This is deliberate: the older `launchctl kickstart -k` only re-execs the process and **reuses launchd's cached environment**, so plist env edits would be silently ignored. If you ever restart the agent by hand, use bootout+bootstrap, not `kickstart -k`:
+
+```bash
+launchctl bootout   gui/$(id -u)/dev.ocp.proxy 2>/dev/null
+launchctl bootstrap gui/$(id -u) ~/Library/LaunchAgents/dev.ocp.proxy.plist
+```
+
+Verify the new value reached the running process:
+
+```bash
+ps -E -p "$(launchctl print gui/$(id -u)/dev.ocp.proxy 2>/dev/null | awk '/pid =/{print $3}')" | tr ' ' '\n' | grep CLAUDE_
+```
+
+On **Linux**, `systemctl --user restart` already re-reads the unit's `EnvironmentFile`, so no special handling is needed.
+
+### Usage shows "unknown"
+
+Usually caused by an expired Claude CLI session. Fix:
+```bash
+claude auth login
+ocp restart
+```
+
+### Startup log warns "OpenClaw registry out of sync"
+
+On boot, OCP compares OpenClaw's registered models against [`models.json`](../models.json) and warns if they drift. Cause: someone (or an OpenClaw upgrade) modified `~/.openclaw/openclaw.json` and removed entries OCP expects. Fix:
+
+```bash
+node ~/ocp/scripts/sync-openclaw.mjs
+```
+
+This is read-only at startup; the warning never blocks the gateway from running.
+
+### A TUI session vanished right after upgrading OCP
+
+If you ran a pre-3.21.1 OCP instance and a post-3.21.1 instance on the same host at the same time during an upgrade, the new instance's one-time boot reap can, once, kill an old-format (`ocp-tui-<8hex>`) live TUI session belonging to the still-running old instance — restart the affected session (`ocp restart` or re-run your TUI turn) and it will come back under the new instance's port-scoped naming.
+
+### OpenClaw shows old models after `ocp update` (v3.10→v3.11 only)
+
+One-time bootstrap quirk for the v3.10.0 → v3.11.0 jump only — the running shell had the old `cmd_update` cached. Run once manually:
+
+```bash
+node ~/ocp/scripts/sync-openclaw.mjs
+openclaw gateway restart   # so OpenClaw re-reads the config
+```
+
+Future `ocp update` invocations sync automatically.
+
+<a id="tui-401"></a>
+### TUI-mode returns a permanent `Please run /login` 401 (re-login doesn't stick)
+
+A long-running TUI-mode host can get stuck returning a permanent 401 (`Please run /login · API Error: 401`) that re-login cannot fix.
+
+**Root cause (two layers):** interactive `claude` **prefers `~/.claude/.credentials.json` over the `CLAUDE_CODE_OAUTH_TOKEN` env var** (this is *unlike* the `-p` path, where the env token wins). So (a) a stale/corrupt `credentials.json` **shadows** the env token — passing the token is not enough on its own; and (b) when claude does use `credentials.json`, its single-use OAuth refresh token can be corrupted (ending up an empty string) by the per-request spawn + `kill-session` teardown racing claude's token rotation. Re-login writes a fresh token, but the next spawn re-corrupts it. Proven live on PI231: *env token passed + broken `credentials.json` present → 401; env token passed + `credentials.json` moved aside → works.*
+
+**Fix:** set `CLAUDE_CODE_OAUTH_TOKEN` on the OCP host and leave `OCP_TUI_HOME` **unset**. OCP then runs the TUI `claude` in a **credential-isolated home** (`$HOME/.ocp-tui/home`) that has **no `credentials.json`** at all, so the env token is the only credential (authoritative — nothing shadows it) and claude never runs the refresh path (so the single-use token can't be corrupted). Then restart — on systemd `daemon-reload`, on launchd `bootout`+`bootstrap`; `kickstart -k` does **not** reload env. Verify the env reached the process and the boot log shows the isolated home:
+
+```bash
+# Linux (systemd): confirm the token is in the service env
+tr '\0' '\n' < /proc/$(pgrep -f server.mjs | head -1)/environ | grep CLAUDE_CODE_OAUTH_TOKEN
+# Boot log should read: TUI-mode: ON home=$HOME/.ocp-tui/home ... auth=env-token (credential-isolated home — no credentials.json)
+```
+
+> If you previously set `OCP_TUI_HOME` to the real home (or any home that contains a `credentials.json`), **unset it** so the credential-isolated default takes effect — otherwise the shadowing `credentials.json` remains in play.
+
+See [Subscription-pool (TUI) mode](tui-mode.md#subscription-pool-tui-mode) and ADR 0007 PR-C / PR-D amendments.

docs/tui-mode.md

@@ -0,0 +1,196 @@
+Part of [OCP](../README.md) — subscription-pool (TUI) mode: serve requests through interactive `claude` so they bill the Pro/Max subscription pool instead of the metered Agent SDK path.
+
+# Subscription-pool (TUI) mode
+
+> **SECURITY — read before enabling.**  
+> TUI-mode is **single-user / single-operator only**. `claude` runs with the OCP process owner's filesystem access regardless of `HOME` setting. If OCP serves multiple users or guest API keys, a guest prompt could exfiltrate files or exhaust the subscription. **Never enable `CLAUDE_TUI_MODE=true` on a multi-user OCP.**
+
+## What it is and why
+
+> **⚠️ Status (as of 2026-07): the billing split below is PAUSED.** Anthropic announced it for 2026-06-15, then paused it on the effective date — *"For now, nothing has changed: Claude Agent SDK, `claude -p`, and third-party app usage still draw from your subscription's usage limits"* ([official help article](https://support.claude.com/en/articles/15036540-use-the-claude-agent-sdk-with-your-claude-plan)). While the pause holds, OCP's default `-p` path bills the subscription and **TUI-mode is a hedge, not a necessity**. The table describes the *announced* regime, kept here because Anthropic says a reworked change will return (with advance notice) — everything in this section is ready to flip on that day.
+
+The announced routing keys `claude` invocations by `cc_entrypoint`:
+
+| Launch method | `cc_entrypoint` | Billing pool (announced regime, currently paused) |
+|---------------|-----------------|-------------|
+| `claude -p` / `--output-format` (OCP default) | `sdk-cli` | Agent SDK credit pool (~$20/mo on Pro) |
+| Interactive `claude` (no flags) | `cli` | Pro/Max subscription pool |
+
+TUI-mode lets OCP serve requests via the interactive path so they bill against the subscription pool under that regime. The response is read from claude's native JSONL session transcript once the turn is complete, then replayed to the caller as a normal OpenAI completion or chunked SSE response.
+
+<a id="tui-entrypoint"></a>
+## Billing-classifier labeling (`OCP_TUI_ENTRYPOINT`)
+
+`OCP_TUI_ENTRYPOINT` (default `cli`) controls how `CLAUDE_CODE_ENTRYPOINT` is set on the spawn
+environment. The default (`cli`) pins the value deterministically — immune to a stray inherited
+env var or a future stdout-redirect bug silently flipping it to `sdk-cli`. This label is honest
+**only** when the spawn is a genuine interactive PTY (tmux pane, no `-p`, stdout not redirected,
+and `tmux new-session` verified to succeed). If you need to observe the raw TTY-derived value, set
+`OCP_TUI_ENTRYPOINT=auto`. See ADR 0007 for the full rationale and governing rule.
+
+## Enabling TUI-mode (opt-in)
+
+```bash
+# Prerequisites
+mkdir -p ~/.ocp-tui/work    # one-time scratch cwd setup
+# tmux must be installed: brew install tmux  /  apt install tmux
+
+# Enable
+export CLAUDE_TUI_MODE=true
+# STRONGLY RECOMMENDED on a TUI host — authenticate via the long-lived OAuth token.
+# With this set (and OCP_TUI_HOME left UNSET), OCP runs the interactive claude in a
+# credential-isolated home ($HOME/.ocp-tui/home, no credentials.json), so the env token
+# is the only credential and is authoritative. This both stops a stale credentials.json
+# from shadowing the token AND ends the refresh-token corruption that caused a permanent
+# "Please run /login" 401 (no credentials file → claude never runs the refresh path).
+# See the auth note below + ADR 0007 PR-D.
+export CLAUDE_CODE_OAUTH_TOKEN=sk-ant-oat01-...
+# Optionally tune:
+export CLAUDE_TUI_WALLCLOCK_MS=180000   # 3 min cap for long Opus turns
+export OCP_TUI_CWD=$HOME/.ocp-tui/work  # default; override if needed
+export OCP_TUI_ENTRYPOINT=cli           # default; use 'auto' to observe TTY-derived value
+# Do NOT set OCP_TUI_HOME for the recommended setup — leaving it unset is what enables
+# the credential-isolated home. Set it only to opt into the legacy symlinked-creds mode.
+```
+
+Then restart OCP. At boot you will see (with the env token set, isolated home auto-selected):
+
+```
+⚠️  TUI-mode ON — single-user only; do NOT enable on a multi-user OCP ...
+  TUI-mode: ON home=/home/user/.ocp-tui/home cwd=/home/user/.ocp-tui/work auth=env-token (credential-isolated home — no credentials.json) wallclock=120000ms maxConcurrent=2
+```
+
+## What changes / what doesn't
+
+- **Callers see no API change.** The response is a normal OpenAI completion object or chunked SSE — identical wire format.
+- **Real streaming is opt-in (`OCP_TUI_STREAM=1`), and off by default.** By default TUI-mode buffers the full response and replays it as chunked SSE — you see a delay, then the complete response. Set `OCP_TUI_STREAM=1` and `stream:true` turns emit real SSE `delta.content` chunks as `claude` renders them, sourced from `claude`'s own `MessageDisplay` hook (byte-faithful raw markdown, on the subscription pool, no `-p`). Two honest caveats: granularity is **block-level** — the hook fires once per rendered block, so a handful of chunks per answer, scaling with length, not token-by-token; and it moves the **first** byte, not the last, so a consumer that must parse a complete reply gains nothing. The transcript stays authoritative: every streamed turn is asserted against it at the end, and a turn whose stream disagrees is **failed rather than served** (watch `tui.streamDivergences` on `/health`). Evidence: [`plans/2026-07-13-tui-latency/streaming-spike.md`](plans/2026-07-13-tui-latency/streaming-spike.md).
+- **Cache and singleflight work normally.** TUI-mode writes the buffered response to the cache on success; cache-hits skip the interactive turn entirely.
+- **The host's `CLAUDE.md` / auto-memory is never injected.** OCP is a proxy — the proxied client (OpenClaw / your IDE) owns its own context and memory. TUI-mode always runs `claude` with `CLAUDE_CODE_DISABLE_CLAUDE_MDS` + `CLAUDE_CODE_DISABLE_AUTO_MEMORY`, so a `CLAUDE.md` on the OCP host can never leak into proxied turns (verified live; see #4). Built-in tool schemas + the interactive system prompt remain (the inherent ~20–35K context floor of interactive mode); MCP is hard-disabled.
+- **Authenticate via `CLAUDE_CODE_OAUTH_TOKEN` in a credential-isolated home (recommended).** tmux does not forward the parent process's env to the pane, so OCP sets the token explicitly on the spawned `claude` when `CLAUDE_CODE_OAUTH_TOKEN` is present. With the env token set and `OCP_TUI_HOME` unset, OCP runs claude in a **credential-isolated home** (`$HOME/.ocp-tui/home`) that has **no `credentials.json`** — so the env token is the only credential and is authoritative, and claude never runs the token-refresh path. This both stops a stale `credentials.json` from shadowing the token and ends the refresh-token corruption behind the permanent `Please run /login · API Error: 401` (full two-layer root cause, live proof, and fix in [Troubleshooting § the permanent TUI-mode 401](troubleshooting.md#tui-401)). Transcripts land under the same isolated home, so the answer-reader is unaffected. Without the env token, claude falls back to the real home's `credentials.json` (byte-for-byte the previous behaviour). (The token is visible in `ps` on the pane command — acceptable for the single-user A-path; the multi-user B-path is refused at boot.) See ADR 0007 PR-C / PR-D amendments.
+- **Stale tmux sessions are reaped.** The pane's `claude` is a child of the tmux server (not OCP), so OCP cannot reap it directly; `claude` zombies can otherwise accumulate as `<defunct>` over a long-running host. OCP reaps them at boot and on a 15-min idle sweep by issuing `tmux kill-server` — but **only when no foreign tmux session remains** (it never disrupts a co-hosted `olp-tui-*` instance). See ADR 0007 PR-C amendment.
+- **Default path unchanged.** Unset `CLAUDE_TUI_MODE` and restart → `callClaude` / `callClaudeStreaming` are used again, byte-for-byte identical to today.
+- **Concurrency is bounded separately.** TUI turns are heavy (per-request cold-boot + long wallclock), so the TUI path has its own limiter — `OCP_TUI_MAX_CONCURRENT` (default `2`), independent of `CLAUDE_MAX_CONCURRENT`. Excess turns queue; a full queue returns a 503. Tune it up only on a host that can run more interactive `claude` sessions at once.
+- **Optional warm pane pool (`OCP_TUI_POOL_SIZE`, default off).** Pre-boots panes so a request skips the cold boot — measured p50 `10.17s` → `6.00s` (−41%). Pooled panes are **single-use** (one turn, then killed and replaced in the background), each carrying its own fresh `--session-id`, so one session still means one exchange and no earlier-turn text can leak into a later answer. They are named `ocp-tui-<port>-p<hex>` and coexist with the reaper by design: the sweep **drains the pool first**, then reaps (so `kill-server` still flushes `<defunct>` zombies), then the pool refills in the background. Drain→reap→resume is synchronous, so no request can land mid-sweep; a request arriving while the pool is still re-booting simply misses it and cold-boots. A live pooled pane is never reaped — **including one that is still booting**, whose tmux session already exists — while an *orphaned* one (left by a previous process generation) still is.
+
+## ⚠️ Latency: TUI mode has a ~6-second floor, and it is immovable
+
+**TUI mode cannot serve real-time or interactive-latency consumers.** This is a hard property of the
+path, stated plainly so you can rule it out before building on it:
+
+| | measured |
+|---|---|
+| **TTFT floor (first token)** | **≈ 6 s** — immovable |
+| cold boot → input bar ready | ~1 s (per request; not the bottleneck) |
+| OCP's own overhead above the CLI | ~4 s (n=1 same-turn decomposition) |
+| direct Anthropic API, same prompt (for scale) | 0.84–1.64 s |
+
+The ~6 s floor is the `claude` CLI itself: it always injects the full Claude Code system prompt plus
+its tool definitions before your prompt, on every turn, no matter what you ask. No flag removes it
+(`--exclude-dynamic-system-prompt-sections` was measured: **no effect** on the floor). Extended
+thinking is *not* the cause — `OCP_TUI_EFFORT` already defaults to `low`, which is what cuts a
+formerly-inherited `xhigh` down to this floor and collapses its variance.
+
+On top of the floor you pay the model's generation time (a function of output length). Progressive
+output is not wired up **yet** (see "No real token streaming" above — it is achievable and planned),
+so today a turn returns as one blob once generation completes. Note that streaming, when it lands,
+will move the *first* byte earlier — it does **not** shorten the turn, and a consumer that needs the
+complete answer gains nothing from it.
+
+**Use TUI mode for**: batch, background, and latency-insensitive work where the subscription pool is
+the point. **Do not use it for**: anything a person is waiting on interactively, or any consumer with
+a sub-5-second budget. Full measurements and methodology:
+[`plans/2026-07-13-tui-latency/`](plans/2026-07-13-tui-latency/).
+
+## Monitoring drift via `/health`
+
+`GET /health` includes a `tui` block so you can poll for a silent billing-pool drift (the top risk under the announced split, if it re-lands — a lost TTY flipping `cc_entrypoint` from `cli` to `sdk-cli` would still return answers but land in the metered pool). The block is **always present** (with `enabled:false` when TUI-mode is off):
+
+```jsonc
+"tui": {
+  "enabled": true,             // CLAUDE_TUI_MODE === "true"
+  "entrypointMode": "cli",     // OCP_TUI_ENTRYPOINT (cli | auto | off)
+  "lastEntrypoint": "cli",     // last cc_entrypoint observed in a transcript, or null
+  "entrypointMismatches": 0,   // count of cli-expected-but-got-other turns — ALERT if this climbs
+  "inflight": 1,               // TUI turns running right now
+  "queued": 0,                 // TUI turns waiting for a concurrency slot
+  "maxConcurrent": 2,          // OCP_TUI_MAX_CONCURRENT
+  "pool": {                    // warm pane pool — null when OCP_TUI_POOL_SIZE=0 (the default)
+    "size": 2,                 // target warm panes (OCP_TUI_POOL_SIZE)
+    "warm": 2,                 // panes ready right now — each is a LIVE idle claude process
+    "booting": 0,              // replacement panes currently pre-booting
+    "model": "claude-sonnet-4-6", // the model being warmed (the most recently requested one)
+    "hits": 12,                // requests served by a warm pane
+    "misses": 1,               // requests that fell back to the cold boot (the 1st is always one)
+    "boots": 14,               // panes successfully pre-booted
+    "bootFailures": 0,         // pre-boots that genuinely never reached the input bar — WATCH this
+    "cancelled": 4,            // in-flight boots OCP killed on purpose (drain / model switch) — not faults
+    "dropped": 8               // panes discarded unused (drain sweep / expired / unhealthy)
+  }
+}
+```
+
+Alert on `entrypointMismatches > 0` (or `lastEntrypoint !== "cli"`): it means a turn drew from the metered Agent SDK pool instead of the subscription. `inflight` / `queued` show how close the TUI path is to its concurrency cap.
+
+With the pool on, `hits` / `misses` is the hit rate (a steady single-model consumer should sit near 100% after the first request), and `warm` is your standing idle-process cost. A climbing `bootFailures` means panes are not reaching their input bar — the pool then degrades safely to the cold path, but latency reverts to the un-pooled numbers. `cancelled` counts boots OCP killed *on purpose* (a drain, a model switch) and is **not** a fault signal — do not alert on it. A steadily climbing `dropped` is likewise normal: the 15-min reap sweep drains and re-boots the pool on every tick so `kill-server` can still flush `<defunct>` zombies.
+
+## Kill-switch
+
+```bash
+unset CLAUDE_TUI_MODE
+# restart OCP
+```
+
+The stream-json path is restored immediately. No other change is needed.
+
+## Operator checklist for the (paused) billing split
+
+> **Status:** the 2026-06-15 split never took effect — Anthropic paused it on the effective date (see the status note at the top of this section). **Nothing needs flipping while the pause holds.** The checklist is retained verbatim as the runbook for if/when a reworked change lands (Anthropic has promised advance notice).
+
+Under the announced regime, every host serving traffic must be flipped to TUI-mode **and** canary-verified before the effective date, or it will bill the metered Agent SDK credit pool instead of the subscription.
+
+- **[Flip/rollback runbook](runbooks/tui-flip-rollback.md)** — how to set `CLAUDE_TUI_MODE=true` on systemd (Linux) and launchd (macOS) hosts. Covers the `daemon-reload` requirement (systemd) and the `bootout`+`bootstrap` cycle requirement (launchd — `launchctl kickstart -k` does not reload plist env).
+- **[615-canary runbook](runbooks/615-canary.md)** — after each flip, run one quiesced request and compare the Agent SDK credit balance before and after. `entrypoint:cli` in the transcript (the `cc_entrypoint` billing classifier) is necessary but not sufficient — only a stable credit balance confirms the subscription pool is being used. Balance check is a manual step (no known programmatic API for the Agent SDK credit pool balance).
+
+## Architecture and design decisions
+
+See [`adr/0007-tui-interactive-mode.md`](adr/0007-tui-interactive-mode.md) for the full rationale, home-strategy options, MCP-disable mechanism, coexistence rules, and the B-path (multi-tenant isolation) roadmap.
+
+## TUI-mode environment variables
+
+The README [Environment Variables](../README.md#environment-variables) table lists these as one-line pointers; the full behaviour of each lives here.
+
+<a id="ocp-tui-stream"></a>
+### `OCP_TUI_STREAM` — real SSE streaming (opt-in)
+
+`OCP_TUI_STREAM` default `0` (off). When `=1`, `stream:true` requests emit **real SSE `delta.content` chunks as `claude` generates them**, instead of buffering the turn and replaying it. Deltas come from `claude`'s own `MessageDisplay` hook (registered with `--settings` on the ordinary interactive spawn — banner-verified to stay on the subscription pool, `· Claude Max`). Granularity is **block-level**, not token-level. The transcript remains authoritative: the streamed text is asserted equal to it at end-of-turn, the auth-banner and truncation gates still run before anything is committed, and only the transcript text is cached. A turn whose stream cannot be reconciled with the transcript is **refused** (SSE error frame, not cached) and counted as `tui.streamDivergences` on `/health`. A total hook failure (e.g. `--settings` stops registering it after a `claude` version bump) is a *different, silent* failure mode — every streamed turn still succeeds, fully buffered, with no divergence and no error — so it is counted separately as `tui.streamZeroDeltaTurns` (streamed turns where the hook fired **zero** times) and logged as `tui_stream_zero_deltas`; watch it alongside `streamDivergences`. Default off — the buffered path is unchanged and remains the stable default. ⚠️ **Tool-using turns:** the transcript keeps only the model's **last** assistant message, so if the model narrates before calling a tool ("I'll check that file…") and that narration exceeds `OCP_TUI_STREAM_HOLDBACK`, it has already been streamed and cannot be retracted — the turn is then **refused** rather than served (measured live: Opus narrated 475 chars before a `Bash` call). If your deployment lets the model use tools (the TUI default, and anything with `OCP_TUI_FULL_TOOLS=1`), either raise `OCP_TUI_STREAM_HOLDBACK` above the typical narration length — the narration then stays held back and is correctly discarded, at the cost of a later first chunk — or leave streaming off. Streaming is best suited to tool-light chat proxying. See ADR 0007 (2026-07-13 amendment).
+
+Two related streaming knobs:
+
+- **`OCP_TUI_STREAM_DIR`** (default `$HOME/.ocp-tui/stream`) — directory holding the static `MessageDisplay` hook script + settings file, and the per-session delta sink (`<session-id>.jsonl`, removed at turn teardown). One sink **per session-id** — this is what keeps concurrent TUI turns (`OCP_TUI_MAX_CONCURRENT` ≥ 2) from interleaving one client's deltas into another's stream.
+- **`OCP_TUI_STREAM_POLL_MS`** (default `100`) — interval at which OCP drains the delta sink. The hook fires at block granularity (seconds apart), so a finer poll buys nothing.
+
+<a id="ocp-tui-stream-holdback"></a>
+### `OCP_TUI_STREAM_HOLDBACK`
+
+`OCP_TUI_STREAM_HOLDBACK` default `100`. (TUI-mode, streaming) Characters withheld before the first chunk reaches the client. Two jobs. (1) It keeps the **auth-banner gate** alive under streaming, via a guarantee with two required halves: (i) nothing is emitted for a message until its trimmed accumulation exceeds 100 chars — past the default banner detector's reach, since real banners are ≤100 chars — and (ii) once a message boundary follows an emit, nothing further is ever emitted for the rest of the turn, and the turn is refused outright. Half (i) alone only covers a turn's first message; half (ii) is what covers an error banner rendered as a *later* message (e.g. after tool-using prose). Raise the holdback if you replace the detector via `CLAUDE_TUI_ERROR_PATTERNS` with patterns that can match longer messages — that only affects half (i); OCP warns at boot if you do. (2) It is the knob for **tool-using turns** — see the `OCP_TUI_STREAM` caveat above. Answers shorter than the holdback are simply delivered whole at end-of-turn, exactly as the buffered path does.
+
+<a id="ocp-tui-pool-size"></a>
+### `OCP_TUI_POOL_SIZE` — warm pane pool
+
+`OCP_TUI_POOL_SIZE` default `0` (off). Number of **pre-booted warm `claude` panes** kept ready, so a request does not pay the cold boot. `0` disables the pool entirely — the request path is then exactly the cold-boot path. Max `4`; an unparseable value disables it rather than guessing. **Measured on a Mac mini (Sonnet 4.6, `--effort low`): end-to-end p50 `10.17s` (n=6, pool off) → `6.00s` (n=12 warm hits) — −4.2 s / −41%** — the pool recovers both the ~1.2 s boot *and* ~2.9 s of post-input-bar init that a pane which has been idle a moment has already finished. **Cost:** each warm pane is a *live idle `claude` process* held whether or not a request ever arrives (peak processes ≈ pool size + `OCP_TUI_MAX_CONCURRENT` + 1 booting replacement) — which is why it is opt-in. Panes are **single-use**: one turn, then killed and replaced in the background. The **first request after start (and after any model switch) is always a cold miss** — the pool warms the most recently requested model, since OCP cannot know which model the next caller wants. See [`plans/2026-07-13-tui-latency/`](plans/2026-07-13-tui-latency/).
+
+<a id="ocp-tui-full-tools"></a>
+### `OCP_TUI_FULL_TOOLS` — full tool surface (single-user only)
+
+`OCP_TUI_FULL_TOOLS` default *(unset)*. (TUI-mode, **single-user only**) When `=1`, grant the interactive session the **same tool surface as the `-p` path** — `--allowedTools` (+ optional `--mcp-config`, read from `CLAUDE_ALLOWED_TOOLS` / `CLAUDE_MCP_CONFIG`) — instead of the default MCP-walled, built-in-tools-only set. Lets a trusted single-operator TUI deployment run a **tool-using / MCP agent** (e.g. an OpenClaw assistant) on the subscription pool. Safe because TUI **refuses to boot under `AUTH_MODE=multi`** (hard exit) — no guest key can ever reach the TUI path, so this gate cannot expose tools to an untrusted caller. (Under `AUTH_MODE=shared` + `OCP_TUI_ALLOW_LAN=1`, anyone holding the single shared key reaches it — that is the existing TUI trust model, unchanged.) Note: `--dangerously-skip-permissions` / `CLAUDE_SKIP_PERMISSIONS` is **not** supported for TUI — claude v2.1.x shows an interactive bypass-acceptance screen in headless tmux that cannot be answered, bricking the pane. Use scratch-home `settings.json` `additionalDirectories` instead. See ADR 0007.
+
+<a id="tui-other-vars"></a>
+### Other TUI-mode variables
+
+- **`OCP_TUI_MAX_CONCURRENT`** (default `2`) — Max concurrent interactive TUI turns. **Independent** of `CLAUDE_MAX_CONCURRENT` (which bounds the `-p`/stream-json path; TUI never uses it). A TUI turn is heavy (per-request cold-boot of tmux+claude + up to `CLAUDE_TUI_WALLCLOCK_MS` wallclock), so the default is low to keep small hosts (e.g. a Pi 4) alive under a burst. Excess turns **queue** (bounded); a full queue yields a 503. See ADR 0007 PR-B amendment.
+- **`OCP_TUI_ENTRYPOINT`** (default `cli`) — Billing-classifier labeling: `cli` (default) pins `cc_entrypoint=cli` deterministically; `auto` lets claude self-classify via TTY detection; `off` leaves the inherited env untouched. Honest only when the spawn is a genuine interactive PTY — see the "Billing-classifier labeling" section above and ADR 0007.
+- **`OCP_TUI_EFFORT`** (default `low`) — Effort level passed to the interactive `claude` as an explicit `--effort` flag: `low` (default), `medium`, `high`, `xhigh`, `max`, or `inherit` to omit the flag (the pre-flag behaviour: the pane inherits a HOME-dependent effort — the operator's `~/.claude/settings.json` `effortLevel` in real-home mode, claude's built-in default in env-token scratch mode). Explicit `low` cuts measured TTFT p50 by ~40% and collapses run-to-run variance ~15× versus an inherited `xhigh` (see [`plans/2026-07-13-tui-latency/`](plans/2026-07-13-tui-latency/)); proxied requests rarely benefit from extended thinking. Banner-verified to stay on the subscription pool (`· Claude Max`). An invalid value logs a warning and falls back to `low`.
+- **`OCP_TUI_HOME`** (default *(auto)*) — `HOME` claude runs under. **When unset, OCP picks it for you:** if `CLAUDE_CODE_OAUTH_TOKEN` is set → a **credential-isolated** scratch home `$HOME/.ocp-tui/home` (no `credentials.json`, env-token auth — **recommended**); if no env token → the operator's real home (legacy shared `credentials.json`). Setting this to an **explicit** path overrides the auto-default. The credential handling at that path still follows the env token: **with** the env token it is credential-free (env-token auth, no `credentials.json` written); **without** the env token (and the path ≠ real home) it uses the legacy symlinked-credentials scratch mode, which carries the credential-fork caveat — see ADR 0007. If you previously set this to the real home (or any home containing a `credentials.json`) and hit a permanent 401, unset it — see [Troubleshooting § the permanent TUI-mode 401](troubleshooting.md#tui-401).
+- **`CLAUDE_TUI_WALLCLOCK_MS`** (default `120000`) — Maximum time in ms to wait for the native transcript to signal turn completion. Increase for long Opus thinking turns.
+- **`OCP_TUI_CWD`** (default `$HOME/.ocp-tui/work`) — Scratch working directory where interactive claude sessions run. Transcripts land under `<HOME>/.claude/projects/<encoded-cwd>/`. Created automatically.
+- **`CLAUDE_CODE_OAUTH_TOKEN`** — the recommended TUI credential; when set (and `OCP_TUI_HOME` unset) it selects the credential-isolated home. Full precedence and the 401 root cause it prevents are in [Troubleshooting § the permanent TUI-mode 401](troubleshooting.md#tui-401).

docs/upgrading.md

@@ -0,0 +1,79 @@
+Part of [OCP](../README.md) — the full upgrade manual (`ocp update` paths, manual flags, rollback, and OpenClaw auto-sync). The README keeps a short stub with the one-liner.
+
+# Upgrading
+
+The simplest path: ask your AI.
+
+  Paste this prompt:
+
+  ```
+  Upgrade my OCP. Run `ocp update` and follow whatever it says.
+  If it tells me to run `claude auth login`, I'll do that.
+  ```
+
+What `ocp update` does:
+
+- **Patch bump** (e.g. `v3.21.0 → v3.21.1`):
+  light path (git pull + npm install + restart).
+- **Cross-minor** (e.g. `v3.18 → v3.22`):
+  full path: pre-flight check, snapshot, `setup.mjs` (with plist env-merge),
+  service restart, post-flight `/health` and `/v1/models` verification.
+- **Old version** (< v3.4.0):
+  fresh-install. Pre-v3.4 lacked admin-key/usage-db, so there is nothing to
+  migrate. Your OAuth token (managed by the Claude Code CLI, not OCP) is
+  preserved; you do not need to re-OAuth unless your token expired
+  separately.
+
+Snapshots are saved to `~/.ocp/upgrade-snapshot-<ISO-ts>/` and never
+auto-deleted. Clean old ones with `rm -rf ~/.ocp/upgrade-snapshot-*` once
+you're confident the upgrade is stable.
+
+## Manual upgrade — same command, no AI
+
+```bash
+ocp update                  # smart-pick path
+ocp update --check          # show available updates, don't apply
+ocp update --dry-run        # preview plan
+ocp update --target v3.13.0 # pin a specific version
+ocp update --rollback --yes # restore most recent snapshot (--yes confirms)
+ocp update --rollback --list      # list snapshots, no mutation
+ocp update --rollback --dry-run   # preview rollback plan
+```
+
+## When upgrade fails
+
+`ocp update` prints a recovery line on failure. To restore from the snapshot:
+
+```bash
+ocp update --rollback --yes   # --yes confirms the destructive restore
+ocp doctor
+```
+
+If `ocp doctor` still reports problems after rollback, open a GitHub issue
+with the snapshot path and the doctor JSON output (`ocp doctor --json`).
+
+## OpenClaw Auto-Sync (v3.11.0+)
+
+Whenever the model list in [`models.json`](../models.json) changes, `ocp update` automatically reconciles your OpenClaw config so the model dropdown stays in sync — no more "I upgraded OCP but my Telegram bot still shows the old models" surprises.
+
+**What gets synced** (and only this — all other config keys are preserved):
+- `models.providers."claude-local".models` in `~/.openclaw/openclaw.json`
+- `agents.defaults.models["claude-local/*"]` aliases
+
+**Safety**:
+- Timestamped backup written before every change: `~/.openclaw/openclaw.json.bak.<ms>`
+- Idempotent — already-in-sync runs are a no-op (no backup, no rewrite)
+- Non-fatal — sync failure does NOT abort `ocp update`; `/v1/models` still works
+- Skips silently if OpenClaw is not installed (`~/.openclaw/openclaw.json` missing)
+
+**Manual trigger** (e.g. after fixing a hand-edited config, or for the one-time v3.10.0→v3.11.0 bootstrap quirk):
+```bash
+node ~/ocp/scripts/sync-openclaw.mjs
+node ~/ocp/scripts/sync-openclaw.mjs --quiet   # silent unless changes
+```
+
+**Opt-out**: `ocp update` only invokes the sync if `node` and `scripts/sync-openclaw.mjs` are both present. Removing the script disables auto-sync; the rest of `ocp update` still works.
+
+**One-time bootstrap caveat (v3.10.0 → v3.11.0 only)**: the first `ocp update` to v3.11.0 runs the *old* `cmd_update` already loaded into your shell, so the new sync hook does NOT fire on this single jump. Run `node ~/ocp/scripts/sync-openclaw.mjs` once manually. Every future update from v3.11.0+ syncs automatically. (Also captured in the README Troubleshooting section as a bootstrap quirk.)
+
+**Other IDEs** (Cline / Aider / Cursor / opencode) query `/v1/models` live, so they pick up new models on the next request — no sync needed. Continue.dev users edit their own `config.json` model id manually.

keys.mjs

@@ -6,26 +6,74 @@ import { join } from "node:path";
 import { mkdirSync, chmodSync } from "node:fs";
 import { homedir } from "node:os";
 
-const OCP_DIR = join(homedir(), ".ocp");
-mkdirSync(OCP_DIR, { recursive: true, mode: 0o700 });
-// Tighten the directory mode in case it already existed with broader permissions.
-try { chmodSync(OCP_DIR, 0o700); } catch { /* ignore EPERM on pre-existing dirs */ }
-const DB_PATH = join(OCP_DIR, "ocp.db");
+// Resolved LAZILY, on first getDb() — not at module top-level. Two reasons, and the second is
+// the bug this fixes:
+//
+//  1. Merely IMPORTING keys.mjs should not, as a side effect, create directories in the
+//     operator's home.
+//  2. OCP_DIR_OVERRIDE exists so the test suite can point the key store at a scratch dir — and
+//     because ESM hoists imports, a top-level `const OCP_DIR = ...` here would be evaluated
+//     BEFORE an importing module's body could set the env var. Eager resolution made the
+//     override unsettable in the one place that needs it. (test-features.mjs carried a comment
+//     claiming it could "set env before the first getDb() call" — it could not, because nothing
+//     here ever read an env var. So `npm test` wrote real, UNREVOKED api_keys rows into the
+//     operator's live ~/.ocp/ocp.db: two per run, unbounded — 737 junk keys against 12 real ones
+//     on the maintainer's host — and two concurrent runs raced one file, which is the ~1-in-6
+//     flake in `listKeys includes quota fields`.)
+//
+// The override is gated on NODE_ENV === "test", and that gate is the ACTUAL guard. An earlier
+// cut of this fix relied on the variable merely having an awkward name — i.e. a naming convention
+// plus a comment — which is precisely the failure mode this whole change exists to indict (a
+// comment describing an intention that nothing enforces). The two-key gate means NEITHER var
+// alone does anything: a stray OCP_DIR_OVERRIDE with no NODE_ENV is inert, and NODE_ENV=test with
+// no override just resolves the default dir.
+//
+// This gate does NOT, by itself, prove a production daemon can't be redirected — an earlier
+// version of this comment overclaimed that ("a production server runs without NODE_ENV, so it
+// CANNOT honor the override no matter how the variable got in"). That is only true while the
+// daemon's env actually lacks NODE_ENV=test, which is an assumption, not something this file can
+// enforce. What makes it hold in the shipped configuration is defense-in-depth in OCP's launchers:
+// the plist/systemd units strip both vars on every (re)install (scripts/lib/plist-merge.mjs
+// NEVER_PRESERVE), and `ocp` restart's manual nohup fallback strips them (`env -u`). So a server
+// OCP itself started cannot carry the test-only redirection. The one residual path is an operator
+// who hand-launches `node server.mjs` with BOTH vars explicitly exported, bypassing every
+// launcher — a case no library-level gate can catch. The loud getDb() log below ("NOT the default
+// ~/.ocp/ocp.db") is the backstop there: a wrong key store is at least never silent (in
+// AUTH_MODE=multi that would otherwise be a total auth outage with nothing on /health to show it).
+function resolveOcpDir() {
+  const override = process.env.NODE_ENV === "test" ? process.env.OCP_DIR_OVERRIDE : null;
+  const dir = override || join(homedir(), ".ocp");
+  mkdirSync(dir, { recursive: true, mode: 0o700 });
+  // Tighten the directory mode in case it already existed with broader permissions.
+  try { chmodSync(dir, 0o700); } catch { /* ignore EPERM on pre-existing dirs */ }
+  return dir;
+}
 
 let db;
+let dbPath;   // resolved on first open, alongside the db handle
 
 export function getDb() {
   if (!db) {
-    db = new DatabaseSync(DB_PATH);
+    dbPath = join(resolveOcpDir(), "ocp.db");
+    // Say which store we opened. Silence was the other half of the bug: a server on the wrong
+    // key store looks exactly like a server on the right one until every request 401s.
+    if (dbPath !== join(homedir(), ".ocp", "ocp.db")) {
+      console.error(`[keys] key store: ${dbPath} (NOT the default ~/.ocp/ocp.db)`);
+    }
+    db = new DatabaseSync(dbPath);
     db.exec("PRAGMA journal_mode = WAL");
     db.exec("PRAGMA foreign_keys = ON");
     initSchema();
     // Tighten mode on the DB file (0600) after creation / first open.
-    try { chmodSync(DB_PATH, 0o600); } catch { /* ignore — same-user access still works */ }
+    try { chmodSync(dbPath, 0o600); } catch { /* ignore — same-user access still works */ }
   }
   return db;
 }
 
+// Which file the key store actually opened. Exported so a test can ASSERT it is not the
+// operator's real db — the bug this replaced was invisible precisely because nothing checked.
+export function getDbPath() { return dbPath; }
+
 function initSchema() {
   db.exec(`
     CREATE TABLE IF NOT EXISTS api_keys (
@@ -307,6 +355,16 @@ export function cacheHash(model, messages, opts = {}) {
   if (opts.temperature != null) h.update(`t:${opts.temperature}`);
   if (opts.max_tokens != null) h.update(`mt:${opts.max_tokens}`);
   if (opts.top_p != null) h.update(`tp:${opts.top_p}`);
+  // #176: fold the server's boot-config epoch into the key, so a config change that shapes
+  // answers (operator system prompt, wrapper text, allowed tools, NO_CONTEXT) invalidates
+  // the persistent cache instead of serving answers composed under the old config. Callers
+  // that omit it (older paths, tests) hash byte-identically to before.
+  if (opts.configEpoch != null) h.update(`ce:${opts.configEpoch}|`);
+  // Structured-output (OpenAI response_format / json_mode) requests must never share a cache slot
+  // with the conversational answer to the same prompt, nor with a different schema — the steering
+  // instruction and validated JSON payload differ. Keying on the detected descriptor isolates them.
+  // Absent for normal requests → hashes are byte-identical to pre-change.
+  if (opts.structured != null) h.update(`s:${JSON.stringify(opts.structured)}`);
   for (const m of messages) {
     h.update(m.role || "");
     h.update(typeof m.content === "string" ? m.content : JSON.stringify(m.content));
@@ -382,11 +440,25 @@ export function getCacheStats() {
 // Per ADR 0005 / spec D4: in-process scope only (single Node process per host).
 const inflightMap = new Map();
 
-export function singleflight(hash, fn) {
+// `retryIf` (optional, audit finding M1): a predicate applied on the FOLLOWER path only.
+// When a follower joins an existing flight and the shared promise rejects with an error for
+// which retryIf(err) is true (in practice: the LEADER's client disconnected while queued —
+// an error that is personal to the leader, not a verdict about the upstream), the follower
+// does NOT inherit that rejection. Instead it re-enters singleflight with its OWN fn: it
+// either becomes the new leader (the map entry is already deleted — see the finally below,
+// which runs before any follower's catch because it is attached upstream of the promise the
+// followers await) or joins a flight another retrying follower just created. The leader's
+// own rejection is never retried here — its error belongs to it (leader path returns the
+// bare promise). Callers that pass no retryIf get the exact pre-M1 share-everything behavior.
+export function singleflight(hash, fn, retryIf) {
   const existing = inflightMap.get(hash);
   if (existing) {
     existing.requesters++;
-    return existing.promise;
+    if (!retryIf) return existing.promise;
+    return existing.promise.catch((err) => {
+      if (!retryIf(err)) throw err;
+      return singleflight(hash, fn, retryIf);
+    });
   }
   // Wrap fn() in Promise.resolve().then() so synchronous throws don't escape.
   const promise = Promise.resolve().then(fn).finally(() => {
@@ -412,5 +484,5 @@ export function findKey(idOrName) {
 }
 
 export function closeDb() {
-  if (db) { db.close(); db = null; }
+  if (db) { db.close(); db = null; dbPath = undefined; }  // clear both — a path to a closed db is a footgun
 }

lib/constants.mjs

@@ -0,0 +1,33 @@
+/**
+ * OCP shared constants — single source of truth.
+ *
+ * Any literal that appears in more than one place across server.mjs, setup.mjs,
+ * scripts/* belongs here so port-drift / URL-drift cascades cannot recur.
+ *
+ * Background: from 2026-05-08 (PR #71 dogfood accident) through 2026-05-13
+ * (v3.16.3) a single hardcoded "3478" in scripts/upgrade.mjs + scripts/doctor.mjs
+ * cascaded into every downstream config write, ultimately taking out the
+ * OpenClaw "大内总管" Telegram agent. See CHANGELOG v3.16.2 and v3.16.3.
+ *
+ * Adding a new constant: prefer ALL_CAPS_SNAKE_CASE. Document the consumers.
+ * If a literal is referenced from a shell script (ocp, ocp-connect, setup.sh)
+ * that can't import .mjs, add a `// keep in sync with lib/constants.mjs` note
+ * at the shell-script reference; CI grep prevents drift.
+ */
+
+// Default TCP port the OCP HTTP proxy listens on. Set by env CLAUDE_PROXY_PORT
+// at runtime; this is the fallback when env is unset.
+// Consumers: server.mjs, setup.mjs, scripts/upgrade.mjs, scripts/doctor.mjs,
+// scripts/sync-openclaw.mjs. Shell scripts ocp / ocp-connect keep the literal
+// "3456" in sync with this value (see CI gate in .github/workflows/alignment.yml).
+export const DEFAULT_PORT = 3456;
+
+// Localhost bind for client-side fetches (curl, health checks).
+export const LOCAL_HOST = "127.0.0.1";
+
+// OpenAI-compatible API base path appended to the proxy URL.
+export const OPENAI_API_BASE = "/v1";
+
+// Convenience: full local URL the OCP proxy listens on by default.
+// scripts that want to probe locally can use this directly.
+export const LOCAL_PROXY_URL = `http://${LOCAL_HOST}:${DEFAULT_PORT}`;

lib/env.mjs

@@ -0,0 +1,29 @@
+// OCP env-var parsing helpers.
+//
+// Fail-closed positive-integer parsing for numeric caps (body size, image
+// byte/count limits). A misconfigured cap must NEVER silently disable a guard:
+// `parseInt("unlimited", 10)` is NaN and `x > NaN` is always false, so a naive
+// parse of CLAUDE_MAX_BODY_SIZE=unlimited would remove the body-size limit
+// entirely (unbounded body → OOM). Likewise CLAUDE_MAX_BODY_SIZE=5MB naively
+// parses to 5 (bytes) and bricks the proxy. So a present-but-invalid value is
+// REJECTED (default kept, caller warns), not accepted. (PR #154 review F3.)
+//
+// Pure (no env access, no IO) so it is unit-testable without a live server.
+
+// Parse `raw` as a strictly-positive base-10 integer of bytes/count (no unit
+// suffix). Returns { value, ok, reason }:
+//   - missing/empty        → { value: def, ok: true }               (use default)
+//   - valid positive int   → { value: n,  ok: true }
+//   - anything else        → { value: def, ok: false, reason }      (fail closed)
+// Rejects: NaN ("unlimited"), non-positive ("0", "-1"), unit-suffixed ("5MB"),
+// and fractional/ambiguous ("20.5", "0x10") values — String(n) !== trimmed catches
+// any input parseInt only partially consumed.
+export function parsePositiveInt(raw, def) {
+  if (raw === undefined || raw === null || raw === "") return { value: def, ok: true };
+  const trimmed = String(raw).trim();
+  const n = parseInt(trimmed, 10);
+  if (!Number.isFinite(n) || n <= 0 || String(n) !== trimmed) {
+    return { value: def, ok: false, reason: "not a strictly-positive integer (bytes/count, no unit suffix)" };
+  }
+  return { value: n, ok: true };
+}

lib/multimodal.mjs

@@ -0,0 +1,278 @@
+// OCP multimodal helpers — OpenAI `image_url` content parts → Anthropic image
+// blocks fed to `claude -p --input-format stream-json`. (issue #110)
+//
+// Class B.1 (OpenAI-compatibility surface). Protocol authority is OpenAI's
+// chat/completions spec — the multimodal `content` parts shape
+// (https://platform.openai.com/docs/guides/vision and
+// https://platform.openai.com/docs/api-reference/chat/create#chat-create-messages,
+// `image_url` part with `image_url.url` = data URI or http(s) URL). Authorized
+// by ADR 0006. This module introduces NO field beyond OpenAI's published shape:
+// the OpenAI-side vocabulary read here is `type:"image_url"` +
+// `image_url:{url, detail?}`; the Anthropic-side vocabulary written here
+// (`type:"image", source:{type:"base64"|"url", ...}`) is the CLI's native
+// stream-json input contract, not an OCP invention.
+//
+// Kept as a pure module (no I/O, no network, no process state) mirroring the
+// lib/*.mjs pattern so it is unit-testable without a live server. server.mjs is
+// the only consumer; it owns spawning, caps configuration, and HTTP status.
+
+// Anthropic vision-supported image media types. A data URI whose media type is
+// outside this set is rejected with a clear 4xx rather than forwarded (the API
+// would reject it anyway; failing early gives a better error).
+export const SUPPORTED_IMAGE_TYPES = new Set([
+  "image/jpeg",
+  "image/png",
+  "image/gif",
+  "image/webp",
+]);
+
+// Default caps. server.mjs overrides these from env; they live here so the
+// pure transform is self-contained and testable.
+export const DEFAULT_MULTIMODAL_OPTS = {
+  allowRemoteUrl: false,        // http(s) image URLs are OFF by default (v1: data URIs only)
+  maxImageBytes: 5 * 1024 * 1024,   // per-image decoded-byte cap
+  maxImages: 20,                // max image parts across the whole request
+  maxTotalImageBytes: 20 * 1024 * 1024, // aggregate decoded-byte cap
+  maxTextChars: Infinity,       // text-char budget (server passes MAX_PROMPT_CHARS); Infinity = no truncation
+};
+
+// Typed error so server.mjs can map to the right HTTP status + OpenAI-shaped
+// error body. `status` is the HTTP code; `type` is the OpenAI error `type`.
+export class MultimodalError extends Error {
+  constructor(code, status, message) {
+    super(message);
+    this.name = "MultimodalError";
+    this.code = code;
+    this.status = status;
+    this.type = "invalid_request_error"; // OpenAI error `type` for 4xx client errors
+  }
+}
+
+// True if any message carries an OpenAI `image_url` content part. Cheap guard so
+// the byte-for-byte text path is only left when an image is genuinely present.
+export function hasImageContent(messages) {
+  if (!Array.isArray(messages)) return false;
+  for (const m of messages) {
+    if (m && Array.isArray(m.content)) {
+      for (const part of m.content) {
+        if (part && part.type === "image_url") return true;
+      }
+    }
+  }
+  return false;
+}
+
+// Extract the URL string from an OpenAI image_url part. Spec form is
+// `{type:"image_url", image_url:{url, detail?}}`; many OpenAI-compatible clients
+// also send `image_url` as a bare string. Accept both (input leniency — no new
+// output field). `detail` (auto|low|high) is OpenAI-only and has no Anthropic
+// analogue, so it is read-and-ignored.
+function imageUrlOf(part) {
+  const iu = part.image_url;
+  if (typeof iu === "string") return iu;
+  if (iu && typeof iu.url === "string") return iu.url;
+  return null;
+}
+
+// Parse a base64 data URI: `data:[<media_type>][;base64],<data>`.
+// Returns { mediaType, data (base64), bytes (decoded size) } or throws MultimodalError.
+function parseDataUri(uri) {
+  const comma = uri.indexOf(",");
+  if (comma === -1) {
+    throw new MultimodalError("invalid_data_uri", 400, "Malformed image data URI (no comma).");
+  }
+  const meta = uri.slice(5, comma); // strip leading "data:"
+  const segs = meta.split(";");
+  const mediaType = (segs[0] || "").trim().toLowerCase();
+  const isBase64 = segs.slice(1).some((s) => s.trim().toLowerCase() === "base64");
+  if (!isBase64) {
+    throw new MultimodalError("invalid_data_uri", 400, "Only base64-encoded image data URIs are supported.");
+  }
+  if (!SUPPORTED_IMAGE_TYPES.has(mediaType)) {
+    throw new MultimodalError(
+      "unsupported_image_type",
+      400,
+      `Unsupported image media type '${mediaType || "(none)"}'. Supported: ${[...SUPPORTED_IMAGE_TYPES].join(", ")}.`
+    );
+  }
+  // Strip incidental whitespace/newlines some encoders insert into data URIs.
+  const data = uri.slice(comma + 1).replace(/\s/g, "");
+  if (data.length === 0 || !/^[A-Za-z0-9+/]+={0,2}$/.test(data)) {
+    throw new MultimodalError("invalid_data_uri", 400, "Image data URI payload is not valid base64.");
+  }
+  // Decoded size from base64 length (minus padding); avoids decoding the buffer
+  // just to measure it.
+  const padding = data.endsWith("==") ? 2 : data.endsWith("=") ? 1 : 0;
+  const bytes = Math.floor((data.length * 3) / 4) - padding;
+  return { mediaType, data, bytes };
+}
+
+// Convert a single OpenAI image_url part to an Anthropic image block, enforcing
+// caps via the mutable `acc` accumulator ({ images, bytes }). Throws MultimodalError.
+function imagePartToBlock(part, opts, acc) {
+  const url = imageUrlOf(part);
+  if (!url) {
+    throw new MultimodalError("invalid_image_url", 400, "image_url part is missing a URL.");
+  }
+
+  acc.images += 1;
+  if (acc.images > opts.maxImages) {
+    throw new MultimodalError("too_many_images", 413, `Too many images in request (max ${opts.maxImages}).`);
+  }
+
+  if (url.startsWith("data:")) {
+    const { mediaType, data, bytes } = parseDataUri(url);
+    if (bytes > opts.maxImageBytes) {
+      throw new MultimodalError("image_too_large", 413, `Image exceeds per-image size limit (${opts.maxImageBytes} bytes).`);
+    }
+    acc.bytes += bytes;
+    if (acc.bytes > opts.maxTotalImageBytes) {
+      throw new MultimodalError("images_too_large", 413, `Total image payload exceeds limit (${opts.maxTotalImageBytes} bytes).`);
+    }
+    return { type: "image", source: { type: "base64", media_type: mediaType, data } };
+  }
+
+  if (/^https?:\/\//i.test(url)) {
+    if (!opts.allowRemoteUrl) {
+      throw new MultimodalError(
+        "remote_url_disabled",
+        400,
+        "Remote image URLs are disabled. Enable CLAUDE_IMAGE_ALLOW_URL=1 to allow http(s) image URLs, or pass the image as a base64 data URI."
+      );
+    }
+    // Passthrough as an Anthropic url-source block. OCP does NOT fetch the URL
+    // itself (no OCP-side SSRF surface); the fetch is performed upstream by the
+    // Anthropic API. Best-effort: unreachable/blocked URLs surface as an API error.
+    return { type: "image", source: { type: "url", url } };
+  }
+
+  throw new MultimodalError("unsupported_url_scheme", 400, "image_url must be a base64 data URI or an http(s) URL.");
+}
+
+// Role prefix mirrors messagesToPrompt()'s text-path labeling so a multi-turn
+// conversation reads the same whether or not it carries images. System messages
+// are handled by the caller via --system-prompt and never reach here.
+function rolePrefix(role) {
+  if (role === "assistant") return "[Assistant] ";
+  return ""; // user / tool / anything else: verbatim, as in the text path
+}
+
+// Build the Anthropic content-block array for a single stream-json user
+// envelope. Mirrors the text path's "collapse the whole conversation into one
+// turn passed via stdin" model (OCP runs stateless, full context per spawn), but
+// preserves image position relative to text and keeps images out of the text
+// char budget entirely. Returns { blocks, stats } or throws MultimodalError.
+export function buildImageBlocks(messages, opts = {}) {
+  const o = { ...DEFAULT_MULTIMODAL_OPTS, ...opts };
+  const blocks = [];
+  const acc = { images: 0, bytes: 0 };
+  let textChars = 0;
+  let firstMessage = true;
+
+  const pushText = (text) => {
+    if (!text) return;
+    blocks.push({ type: "text", text });
+    textChars += text.length;
+  };
+
+  for (const m of messages) {
+    const prefix = rolePrefix(m.role);
+    // Separate messages with a blank line, matching messagesToPrompt's "\n\n" join.
+    const sep = firstMessage ? "" : "\n\n";
+    firstMessage = false;
+    let prefixEmitted = false;
+    const emitPrefixWith = (t) => {
+      if (prefixEmitted) return t;
+      prefixEmitted = true;
+      return sep + prefix + t;
+    };
+
+    if (typeof m.content === "string") {
+      pushText(emitPrefixWith(m.content));
+      continue;
+    }
+    if (!Array.isArray(m.content)) {
+      // null / object content: mirror contentToText's fallback.
+      const t = m.content == null ? "" : JSON.stringify(m.content);
+      pushText(emitPrefixWith(t));
+      continue;
+    }
+
+    for (const part of m.content) {
+      if (part && part.type === "text" && typeof part.text === "string") {
+        pushText(emitPrefixWith(part.text));
+      } else if (part && part.type === "image_url") {
+        // Ensure the role prefix isn't lost when a message leads with an image.
+        if (!prefixEmitted && prefix) pushText(emitPrefixWith(""));
+        blocks.push(imagePartToBlock(part, o, acc));
+      } else {
+        // audio / file / unknown parts: preserve the existing placeholder
+        // behavior (issue #110) — deferred to a future version.
+        pushText(emitPrefixWith("[non-text content omitted]"));
+      }
+    }
+  }
+
+  // Defensive: a stream-json user turn must have at least one content block.
+  if (blocks.length === 0) blocks.push({ type: "text", text: "" });
+
+  // Enforce the text-char budget (PR #154 review F2). Without this, attaching a
+  // single tiny image would let unbounded text bypass the gateway's runaway-context
+  // guard entirely — messagesToPrompt truncates the text path, so the multimodal
+  // path must too. Image blocks are preserved and are NOT counted (they are bounded
+  // by the byte/count caps above).
+  const budgeted = enforceTextBudget(blocks, o.maxTextChars);
+
+  return {
+    blocks: budgeted.blocks,
+    stats: {
+      imageCount: acc.images,
+      totalImageBytes: acc.bytes,
+      textChars: budgeted.textChars,
+      originalTextChars: budgeted.originalTextChars,
+      truncated: budgeted.truncated,
+    },
+  };
+}
+
+// Enforce a text-char budget over already-built content blocks, mirroring
+// messagesToPrompt's "keep the tail, drop the oldest" truncation. Image blocks are
+// preserved in place; only text blocks count against the budget. A truncation note
+// is prepended when anything is dropped. Returns
+// { blocks, truncated, originalTextChars, textChars }.
+function enforceTextBudget(blocks, maxTextChars) {
+  let originalTextChars = 0;
+  for (const b of blocks) if (b.type === "text") originalTextChars += b.text.length;
+  if (!(maxTextChars > 0) || originalTextChars <= maxTextChars) {
+    return { blocks, truncated: false, originalTextChars, textChars: originalTextChars };
+  }
+  // Keep the most recent text up to the budget; trim within the boundary block and
+  // drop older text blocks. Non-text (image) blocks always survive, in order.
+  let budget = maxTextChars;
+  const out = [];
+  for (let i = blocks.length - 1; i >= 0; i--) {
+    const b = blocks[i];
+    if (b.type !== "text") { out.unshift(b); continue; }
+    if (budget <= 0) continue; // older text fully dropped
+    if (b.text.length <= budget) {
+      out.unshift(b);
+      budget -= b.text.length;
+    } else {
+      out.unshift({ type: "text", text: b.text.slice(b.text.length - budget) });
+      budget = 0;
+    }
+  }
+  out.unshift({ type: "text", text: "[System] Note: older text content was truncated to fit the context limit." });
+  let textChars = 0;
+  for (const b of out) if (b.type === "text") textChars += b.text.length;
+  return { blocks: out, truncated: true, originalTextChars, textChars };
+}
+
+// Serialize the non-system conversation to a single newline-terminated
+// stream-json user message for `claude -p --input-format stream-json` stdin.
+// Returns { payload, stats } or throws MultimodalError.
+export function buildStreamJsonInput(messages, opts = {}) {
+  const { blocks, stats } = buildImageBlocks(messages, opts);
+  const envelope = { type: "user", message: { role: "user", content: blocks } };
+  return { payload: JSON.stringify(envelope) + "\n", stats };
+}

lib/net.mjs

@@ -0,0 +1,9 @@
+// OCP network helpers — shared so server.mjs and tests use one definition. (issue #125)
+
+// A bind address is "loopback" only if it cannot be reached from another host.
+// Any other address (0.0.0.0, ::, a concrete LAN/Tailscale IP, etc.) is
+// network-exposed and must trigger the TUI LAN gate.
+export function isLoopbackBind(addr) {
+  return addr === "127.0.0.1" || addr === "::1" || addr === "localhost" ||
+         addr === "::ffff:127.0.0.1" || /^127\./.test(addr);
+}

lib/prompt.mjs

@@ -0,0 +1,89 @@
+// lib/prompt.mjs — pure operator-append step for the system prompt.
+//
+// Extracted so the rule is unit-testable (the suite never imports server.mjs — it
+// boots a listener). server.mjs composes wrapper + client system messages exactly as
+// before, then passes the result through this. With CLAUDE_SYSTEM_PROMPT unset the
+// return is the INPUT STRING UNCHANGED — the default path stays byte-for-byte
+// identical, which is the repo's bar for touching a request-shaping function.
+//
+// The operator prompt goes LAST deliberately: a server-wide directive ("answer in
+// Chinese") should read as the final instruction, not something a client system
+// message overrides by coming later. Whitespace-only values are treated as unset —
+// a stray space in a service unit's Environment= line must not inject "\n\n " into
+// every request.
+export function appendOperatorPrompt(base, operatorAppend) {
+  const op = typeof operatorAppend === "string" ? operatorAppend.trim() : "";
+  return op ? `${base}\n\n${op}` : base;
+}
+
+// Derive the default prompt-char budget from the models.json SPOT (ADR 0009).
+//
+// The old default was a hand-set constant (150000 chars ≈ 37.5k English tokens) from the
+// 200k-window era — silently far below what the advertised contextWindow promises. Instead
+// of picking a new constant that will also rot, the default now FOLLOWS the SPOT:
+//
+//   budget = max(models[].contextWindow) × charsPerToken
+//
+// charsPerToken = 3 is deliberately conservative: English runs ~4 chars/token, CJK ~1–1.5.
+// At ×3, a 200k-token window yields 600,000 chars — full window for English, and CJK text
+// hits the model's real window at roughly the same point the cap fires, so we truncate
+// (graceful, tail-first) rather than let the upstream reject the request outright.
+//
+// The floor guards the degenerate cases (empty/missing models[], absent contextWindow):
+// fall back to the historical constant rather than 0 — a zero budget would truncate every
+// request to nothing, which is fail-OPEN in the "serve garbage" sense. CLAUDE_MAX_PROMPT_CHARS
+// remains an absolute operator override at the call site (server.mjs); this function is only
+// the unset-env default.
+export function derivePromptCharBudget(models, { charsPerToken = 3, floor = 150000 } = {}) {
+  const windows = (Array.isArray(models) ? models : [])
+    .map(m => m?.contextWindow)
+    .filter(w => Number.isFinite(w) && w > 0);
+  if (windows.length === 0) return floor;
+  return Math.max(floor, Math.max(...windows) * charsPerToken);
+}
+
+// Resolve the effective budget from the env var + SPOT. TRUTHINESS (not != null) on the env
+// value deliberately: an EMPTY value ("CLAUDE_MAX_PROMPT_CHARS=" in a systemd EnvironmentFile
+// or .env) must mean "use the default" — exactly the old `parseInt(env || "150000")` contract.
+// Treating "" as explicit gives parseInt("") = NaN, and a NaN cap silently DISABLES the
+// runaway-context guard while injecting a false "[System] Note: 0 older messages were
+// truncated" line into every prompt (caught in PR #179 review). Non-empty garbage still
+// parses to NaN — the pre-existing class, slated for parseIntEnv routing in PR #154.
+export function resolvePromptCharBudget(rawEnv, models, opts) {
+  return rawEnv ? parseInt(rawEnv, 10) : derivePromptCharBudget(models, opts);
+}
+
+// OCP_LOCAL_TOOLS system-prompt wrapper selection (pure).
+//
+// OCP's `-p` path prepends a fixed wrapper to every request's system prompt. The DEFAULT wrapper
+// tells the model it has NO local filesystem/shell/env access — the right posture for a shared or
+// multi-tenant gateway. But a single-user, loopback-bound instance (e.g. an OpenClaw agent talking
+// to its own local OCP) DOES legitimately have tools — the `-p` path already passes `--allowedTools`
+// and the CLI's built-in tools are available — so the default wrapper actively gaslights the model
+// into refusing to use tools it holds. `OCP_LOCAL_TOOLS=1` swaps in a positive wrapper for that case.
+//
+// This does NOT expand the tool surface: tools are governed solely by `--allowedTools` /
+// `--disallowedTools` (multi-tenant mode `--disallowedTools` the whole FS surface regardless of the
+// wrapper). It only changes the PROMPT the operator's own model reads. Pure so it is unit-testable.
+export function selectPromptWrapper(localToolsEnabled, negativeWrapper, positiveWrapper) {
+  return localToolsEnabled ? positiveWrapper : negativeWrapper;
+}
+
+// Boot-time safety gate for OCP_LOCAL_TOOLS, mirroring the OCP_TUI_FULL_TOOLS model (ADR 0007): a
+// positive "you may use local tools" wrapper must never reach an untrusted caller. Returns a fatal
+// message string when the flag is enabled in an unsafe deployment, or null when it is safe/disabled.
+// Fail-closed: any of multi-tenant auth, a non-loopback bind, or an anonymous key is refused. Pure —
+// the caller does the process.exit so this stays testable.
+export function localToolsSafetyError({ enabled, authMode, loopbackBind, anonymousKey }) {
+  if (!enabled) return null;
+  if (authMode === "multi") {
+    return "OCP_LOCAL_TOOLS=1 is incompatible with CLAUDE_AUTH_MODE=multi — a guest/anonymous prompt would be told it may drive the operator's filesystem/shell. Single-user only.";
+  }
+  if (!loopbackBind) {
+    return "OCP_LOCAL_TOOLS=1 requires a loopback bind (127.0.0.1/::1) — a network-exposed positive-tools wrapper could reach an untrusted peer. Bind to loopback, or leave OCP_LOCAL_TOOLS off.";
+  }
+  if (anonymousKey) {
+    return "OCP_LOCAL_TOOLS=1 is unsafe with PROXY_ANONYMOUS_KEY set — anonymous callers could reach the local-tools-enabled model without a named key. Remove PROXY_ANONYMOUS_KEY, or leave OCP_LOCAL_TOOLS off.";
+  }
+  return null;
+}

lib/spawn-auth.mjs

@@ -0,0 +1,67 @@
+// Pure, dependency-injected primitives for the `-p` spawn-token resolution + HOME-isolation
+// layer. Extracted from server.mjs (findings F3 / F5 / F6, 2026-07-07) so the concurrency,
+// caching and expiry logic is unit-testable WITHOUT booting the server or mocking execFileSync /
+// child_process.spawn / fs. server.mjs owns all I/O (macOS keychain exec, process spawn, fs);
+// this module owns only pure decision logic.
+//
+// ALIGNMENT NOTE: none of this touches the OAuth wire machinery (no endpoint / header / body).
+// OCP still NEVER performs a refresh_token grant itself — these helpers only READ + GATE a token
+// that some other process (the operator's real claude, or a spawned claude under the real HOME)
+// refreshes. That property is load-bearing (issue #112) and preserved.
+
+// Promise-chain mutex. `acquire()` resolves to a `release()` fn; the NEXT `acquire()` does not
+// resolve until the current holder calls its `release()`. Serializes async critical sections
+// without busy-waiting. release() is idempotent.
+export function createSerialMutex() {
+  let tail = Promise.resolve();
+  return {
+    acquire() {
+      let release;
+      const gate = new Promise((r) => { release = r; });
+      const prev = tail;
+      tail = tail.then(() => gate);
+      // Hand the caller its release fn only after the previous holder has released.
+      return prev.then(() => {
+        let released = false;
+        return function releaseMutex() { if (!released) { released = true; release(); } };
+      });
+    },
+  };
+}
+
+// Short-TTL memo. `get(produce, now)` returns the cached value while `now - storedAt < ttlMs`,
+// otherwise calls `produce()` and re-stores. A miss that produces null/undefined is STILL stored
+// (so a genuinely-absent source is not re-probed on every call within the TTL window). `now` is
+// injectable for testing.
+export function createTtlCache({ ttlMs }) {
+  let value;
+  let at = -Infinity;
+  let has = false;
+  return {
+    get(produce, now = Date.now()) {
+      if (has && now - at < ttlMs) return value;
+      value = produce();
+      at = now;
+      has = true;
+      return value;
+    },
+    clear() { has = false; value = undefined; at = -Infinity; },
+  };
+}
+
+// Pure expiry gate. Returns true when `creds` carries a known expiry that is at/within `bufferMs`
+// of `now`. Creds WITHOUT `expiresAt` (e.g. long-lived env tokens) are never treated as expiring.
+// This gate is applied to the CACHED creds on EVERY use — which is precisely why a short-TTL
+// keychain cache (createTtlCache) cannot reintroduce the #146 forever-stale-token regression: the
+// cache bounds how often we re-READ the keychain, but the expiry decision is recomputed per use.
+export function isTokenExpiring(creds, now = Date.now(), bufferMs = 300000) {
+  return !!(creds && creds.expiresAt && now + bufferMs >= creds.expiresAt);
+}
+
+// Order candidate keychain labels so the last-known-good label is tried first (avoids the
+// wrong-label miss that doubles the `security` exec count on the hot path). Pure: performs no
+// read. Returns a fresh array; input is not mutated.
+export function orderLabelsLastGoodFirst(labels, lastGood) {
+  if (!lastGood || !labels.includes(lastGood)) return labels.slice();
+  return [lastGood, ...labels.filter((l) => l !== lastGood)];
+}

lib/structured-output.mjs

@@ -0,0 +1,318 @@
+// ── OpenAI Structured Outputs (response_format) — pure helpers ───────────────
+//
+// OCP's `/v1/chat/completions` (Class B.1, ADR 0006) advertises OpenAI compatibility but forwards
+// to `claude -p`, which has no native `response_format`. Asked for JSON, the coding-assistant CLI
+// typically replies with prose, a Markdown table, or a ```json fenced block — none of which is
+// `JSON.parse`-able. These helpers implement the OpenAI `response_format` contract on top of that:
+// detect the request, build a strict JSON-only steering instruction, then extract and validate the
+// JSON the model returns. All functions here are pure (no I/O) so they are unit-tested directly;
+// the retry loop that calls the model lives in server.mjs (runStructuredCompletion).
+//
+// Spec authority (B.1, ADR 0006): OpenAI chat/completions `response_format`
+//   https://platform.openai.com/docs/api-reference/chat/create#chat-create-response_format
+// No field or behaviour beyond that published shape is introduced.
+
+export class StructuredOutputError extends Error {
+  constructor(reason, raw) {
+    super(`structured output could not be produced: ${reason}`);
+    this.name = "StructuredOutputError";
+    this.reason = reason;
+    this.raw = raw;
+  }
+}
+
+// Fail-closed parse of the OCP_STRUCTURED_MAX_ATTEMPTS retry cap. `Math.max(1, parseInt("abc",10))`
+// === `Math.max(1, NaN)` === NaN, and a retry loop bounded by `attempt < NaN` never runs → 0 spawns,
+// every structured request silently refuses. So any non-integer / non-finite / <1 value keeps the
+// documented default instead (and warns), rather than bricking the feature. (PR #153 review round 2.)
+export function resolveMaxAttempts(raw, { fallback = 3, warn } = {}) {
+  if (raw === undefined || raw === null || raw === "") return fallback;
+  const n = parseInt(raw, 10);
+  if (!Number.isFinite(n) || n < 1) {
+    if (typeof warn === "function") {
+      warn(`Ignoring invalid OCP_STRUCTURED_MAX_ATTEMPTS="${raw}" (want integer >= 1); using default ${fallback}.`);
+    }
+    return fallback;
+  }
+  return n;
+}
+
+// Returns { mode: "schema", schema, name?, strict } | { mode: "json_object" } | null.
+// Supports the OpenAI shapes: response_format:{type:"json_schema",json_schema:{schema,strict,name}}
+// and response_format:{type:"json_object"}, a lenient response_format:{schema} fallback, and the
+// widely-used (non-standard) top-level `json_mode: true` flag as a json_object alias.
+export function detectStructuredOutput(parsed) {
+  const rf = parsed?.response_format;
+  if (rf && typeof rf === "object") {
+    if (rf.type === "json_schema") {
+      const js = (rf.json_schema && typeof rf.json_schema === "object") ? rf.json_schema : {};
+      const schema = js.schema || rf.schema || null;
+      return { mode: "schema", schema, name: js.name, strict: js.strict === true };
+    }
+    if (rf.type === "json_object") return { mode: "json_object" };
+    if (rf.schema && typeof rf.schema === "object") {
+      return { mode: "schema", schema: rf.schema, strict: rf.strict === true };
+    }
+  }
+  // Non-standard convenience alias honored by several OpenAI-compatible clients.
+  if (parsed?.json_mode === true) return { mode: "json_object" };
+  return null;
+}
+
+export function jsonTypeOf(v) {
+  if (v === null) return "null";
+  if (Array.isArray(v)) return "array";
+  return typeof v; // object | string | number | boolean
+}
+
+export function jsonTypeMatches(t, value) {
+  switch (t) {
+    case "string": return typeof value === "string";
+    case "number": return typeof value === "number" && Number.isFinite(value);
+    case "integer": return typeof value === "number" && Number.isInteger(value);
+    case "boolean": return typeof value === "boolean";
+    case "object": return value !== null && typeof value === "object" && !Array.isArray(value);
+    case "array": return Array.isArray(value);
+    case "null": return value === null;
+    default: return true; // unknown type keyword → do not fail on it
+  }
+}
+
+const jsonDeepEqual = (a, b) => JSON.stringify(a) === JSON.stringify(b);
+
+// Resolve a local JSON-Pointer `$ref` (e.g. "#/$defs/Step" or "#/definitions/Step") against the
+// document root. Only same-document refs are supported (that is all the OpenAI SDK emits); a remote
+// or unresolvable ref returns null and the caller skips validation for it rather than failing.
+function resolveRef(ref, root) {
+  if (typeof ref !== "string" || !ref.startsWith("#/") || !root) return null;
+  const parts = ref.slice(2).split("/").map(p => p.replace(/~1/g, "/").replace(/~0/g, "~"));
+  let cur = root;
+  for (const p of parts) {
+    if (cur && typeof cur === "object" && Object.prototype.hasOwnProperty.call(cur, p)) cur = cur[p];
+    else return null;
+  }
+  return (cur && typeof cur === "object") ? cur : null;
+}
+
+// Minimal JSON-Schema validator: covers the subset OpenAI structured outputs use — type (incl.
+// arrays of types / integer), required, properties, additionalProperties (no invented keys), items
+// (list + tuple), enum, const, nullability (type:["x","null"] or nullable:true), min/maxItems, and
+// $ref/$defs + allOf/anyOf/oneOf composition (which the official OpenAI SDK emits heavily via
+// zodResponseFormat / client.beta.chat.completions.parse). `root` carries the top-level schema so
+// same-document $refs resolve. Returns error strings ([] = valid).
+// `refChain` tracks the $ref pointers resolved on the CURRENT path WITHOUT consuming data (a $ref
+// hop, or an allOf/anyOf/oneOf branch, all re-validate the same value). A pointer reappearing on
+// that chain is a pure ref→ref (or ref→composition→ref) cycle that recurses forever independent of
+// the data — we fail closed on it. Data-consuming recursion (properties/items/additionalProperties)
+// deliberately resets the chain (default []): it always terminates because a JSON value is a finite
+// tree, so a legitimately recursive schema (Node→child:Node) must NOT be flagged as a cycle. A depth
+// cap backstops any threading mistake. (PR #153 review round 2, cyclic-$ref blocker.)
+const REF_DEPTH_CAP = 512;
+export function validateJsonSchema(value, schema, path = "$", strict = false, root = schema, refChain = []) {
+  const errors = [];
+  if (!schema || typeof schema !== "object") return errors;
+  if (refChain.length > REF_DEPTH_CAP) { // defensive backstop; refChain cycle-check below is primary
+    errors.push(`${path}: $ref resolution too deep (possible cycle)`);
+    return errors;
+  }
+
+  // $ref: resolve against the document root ($defs / definitions) and validate the target. Without
+  // this a nested {$ref:"#/$defs/Step"} presents as {no type, no properties} — and under strict that
+  // used to wrongly reject every real key as "additional property not allowed" (the flagship OpenAI
+  // SDK shape). Sibling keywords alongside $ref (rare) are merged over the resolved target.
+  if (typeof schema.$ref === "string") {
+    if (refChain.includes(schema.$ref)) { // cyclic $ref (a→b→a, or self a→a) — fail closed.
+      errors.push(`${path}: cyclic $ref detected (${schema.$ref})`);
+      return errors;
+    }
+    const resolved = resolveRef(schema.$ref, root);
+    if (!resolved) return errors; // unresolvable ref → cannot validate; do not fail
+    const { $ref, ...siblings } = schema;
+    return validateJsonSchema(value, { ...resolved, ...siblings }, path, strict, root, [...refChain, schema.$ref]);
+  }
+
+  // Composition. allOf: every branch must pass. anyOf: at least one. oneOf: exactly one.
+  // These re-validate the SAME value → thread refChain so a ref cycle routed through a branch is caught.
+  if (Array.isArray(schema.allOf)) {
+    for (const sub of schema.allOf) errors.push(...validateJsonSchema(value, sub, path, strict, root, refChain));
+  }
+  if (Array.isArray(schema.anyOf)) {
+    if (!schema.anyOf.some(sub => validateJsonSchema(value, sub, path, strict, root, refChain).length === 0)) {
+      errors.push(`${path}: does not match any of the allowed schemas (anyOf)`);
+    }
+  }
+  if (Array.isArray(schema.oneOf)) {
+    const matches = schema.oneOf.filter(sub => validateJsonSchema(value, sub, path, strict, root, refChain).length === 0).length;
+    if (matches !== 1) errors.push(`${path}: must match exactly one allowed schema (oneOf), matched ${matches}`);
+  }
+
+  // Nullability takes precedence: a null value is valid whenever the schema permits null (its type
+  // union includes "null", or nullable:true), regardless of enum/const. This mirrors OpenAI
+  // structured-output behaviour — nullable fields accept null even when a bare enum (as generated by
+  // Home Assistant's extended_openai_conversation) omits null from its value list.
+  const allowsNull = schema.nullable === true
+    || (Array.isArray(schema.type) ? schema.type.includes("null") : schema.type === "null");
+  if (value === null && allowsNull) return errors;
+
+  if (Array.isArray(schema.enum) && !schema.enum.some(e => jsonDeepEqual(e, value))) {
+    errors.push(`${path}: not one of the allowed enum values`);
+  }
+  if ("const" in schema && !jsonDeepEqual(schema.const, value)) {
+    errors.push(`${path}: does not equal the required const value`);
+  }
+
+  if (schema.type !== undefined) {
+    const types = Array.isArray(schema.type) ? schema.type : [schema.type];
+    const nullable = schema.nullable === true || types.includes("null");
+    const ok = types.some(t => jsonTypeMatches(t, value)) || (value === null && nullable);
+    if (!ok) {
+      errors.push(`${path}: expected ${types.join("|")}${schema.nullable ? "|null" : ""}, got ${jsonTypeOf(value)}`);
+      return errors; // type mismatch — deeper checks are meaningless
+    }
+  }
+  if (value === null) return errors;
+
+  const vt = jsonTypeOf(value);
+  if (vt === "object") {
+    const props = schema.properties || {};
+    for (const r of (schema.required || [])) {
+      if (!Object.prototype.hasOwnProperty.call(value, r)) errors.push(`${path}.${r}: required property missing`);
+    }
+    const addl = schema.additionalProperties;
+    // Only treat strict as implying "no additional properties" when this object actually declares
+    // its own `properties` and is NOT a composite (allOf/anyOf/oneOf put the real keys in sub-schemas,
+    // which are validated separately above). Inferring closure from an EMPTY properties map — the
+    // shape an unresolved $ref or a pure-composition node presents — would reject every real key.
+    // An explicit additionalProperties:false is always honoured. (PR #153 review, finding 1.)
+    const isComposite = Array.isArray(schema.allOf) || Array.isArray(schema.anyOf) || Array.isArray(schema.oneOf);
+    const noExtra = addl === false || (strict && addl === undefined && Object.keys(props).length > 0 && !isComposite);
+    for (const k of Object.keys(value)) {
+      if (props[k]) {
+        errors.push(...validateJsonSchema(value[k], props[k], `${path}.${k}`, strict, root));
+      } else if (isComposite) {
+        // key may be defined in an allOf/anyOf/oneOf branch — already validated there; don't reject.
+      } else if (noExtra) {
+        errors.push(`${path}.${k}: additional property not allowed`);
+      } else if (addl && typeof addl === "object") {
+        errors.push(...validateJsonSchema(value[k], addl, `${path}.${k}`, strict, root));
+      }
+    }
+  } else if (vt === "array" && schema.items) {
+    if (Array.isArray(schema.items)) {
+      schema.items.forEach((s, i) => { if (i < value.length) errors.push(...validateJsonSchema(value[i], s, `${path}[${i}]`, strict, root)); });
+    } else {
+      value.forEach((item, i) => errors.push(...validateJsonSchema(item, schema.items, `${path}[${i}]`, strict, root)));
+    }
+    if (typeof schema.minItems === "number" && value.length < schema.minItems) errors.push(`${path}: fewer items than minItems ${schema.minItems}`);
+    if (typeof schema.maxItems === "number" && value.length > schema.maxItems) errors.push(`${path}: more items than maxItems ${schema.maxItems}`);
+  }
+  return errors;
+}
+
+// Crash-safe façade over validateJsonSchema (issue #181). The validator recurses on the DATA's
+// nesting depth (properties/items/additionalProperties), which the REF_DEPTH_CAP does NOT bound —
+// only the ref-chain is. A model reply nested ~2000 levels deep therefore overflowed the stack with
+// a RangeError, which the request handler caught as a generic HTTP 500 instead of the spec-correct
+// `refusal`. This wrapper converts ANY throw (the deep-data RangeError, or any future recursion
+// hazard) into a single validation error, so the structured-output retry loop treats a pathological
+// reply as "did not validate" → refusal — never a 500, never a crash. A well-formed reply is
+// unaffected: the inner validator returns and this just passes its errors through.
+export function validateJsonSchemaSafe(value, schema, path = "$", strict = false, root = schema) {
+  try {
+    return validateJsonSchema(value, schema, path, strict, root);
+  } catch (e) {
+    // Catch ONLY the deep-nesting stack overflow (the #181 vector) and turn it into a validation
+    // miss → retry → refusal, never a 500. Any OTHER throw is a genuine bug: re-throw it so it
+    // surfaces at error level instead of being silently masked as "did not validate" (reviewer
+    // finding — a catch-all would hide a future TypeError behind a warn-level structured_retry).
+    if (e instanceof RangeError) return [`${path}: schema validation aborted (value nesting too deep)`];
+    throw e;
+  }
+}
+
+function tryJsonParse(s) {
+  try { return { ok: true, value: JSON.parse(s) }; } catch { return { ok: false }; }
+}
+
+// Find the next brace-balanced JSON span at or after `from`. String-aware: brackets inside quoted
+// strings are ignored. Returns { text, start, end } for the first complete top-level span, or null.
+function balancedSlice(s, from) {
+  let start = -1;
+  for (let i = from; i < s.length; i++) { if (s[i] === "{" || s[i] === "[") { start = i; break; } }
+  if (start === -1) return null;
+  let depth = 0, inStr = false, esc = false;
+  for (let i = start; i < s.length; i++) {
+    const c = s[i];
+    if (inStr) {
+      if (esc) esc = false;
+      else if (c === "\\") esc = true;
+      else if (c === '"') inStr = false;
+      continue;
+    }
+    if (c === '"') { inStr = true; continue; }
+    if (c === "{" || c === "[") depth++;
+    else if (c === "}" || c === "]") { depth--; if (depth === 0) return { text: s.slice(start, i + 1), start, end: i }; }
+  }
+  return null;
+}
+
+// Extract a single JSON value from model text. Two modes (PR #153 review, finding 2):
+//
+//   opts.whole === true  (json_object mode): the ENTIRE reply, after trimming and stripping one code
+//     fence, must parse as a single JSON value. json_object has no schema to validate against, so
+//     this whole-reply parse is its ONLY guard — a reply like `I can't. The schema is {"type":"object"}`
+//     must NOT parse-and-serve the embedded object as if it were the answer.
+//
+//   default (schema mode): prose-wrapped JSON is tolerated (models often add a sentence), BUT a reply
+//     containing MORE THAN ONE top-level JSON value is rejected rather than silently serving the first
+//     — "Schema: {...}\n\nAnswer: {...}" or "Option A: {...} Option B: {...}" is ambiguous, not an
+//     answer. The extracted value is still schema-validated by the caller.
+//
+// Returns { ok:true, value } | { ok:false, reason? }.
+export function extractJsonPayload(text, opts = {}) {
+  if (typeof text !== "string") return { ok: false };
+  let s = text.trim();
+  const fence = s.match(/^```(?:json)?\s*([\s\S]*?)\s*```$/i);
+  if (fence) s = fence[1].trim();
+
+  if (opts.whole) {
+    const whole = tryJsonParse(s);
+    return whole.ok ? whole : { ok: false, reason: "reply was not a single JSON value" };
+  }
+
+  const direct = tryJsonParse(s);
+  if (direct.ok) return direct;
+
+  const first = balancedSlice(s, 0);
+  if (!first) return { ok: false };
+  const parsedFirst = tryJsonParse(first.text);
+  if (!parsedFirst.ok) return { ok: false };
+
+  // Reject ambiguity: a second parseable top-level JSON value means we cannot know which is the answer.
+  const second = balancedSlice(s, first.end + 1);
+  if (second && tryJsonParse(second.text).ok) {
+    return { ok: false, reason: "reply contained more than one JSON value" };
+  }
+  return parsedFirst;
+}
+
+// The strict JSON-only system instruction appended to the request (attempt 0), escalated with the
+// prior failure reason on retries.
+export function structuredSystemInstruction(structured, attempt, lastErr) {
+  const schemaBlock = (structured.mode === "schema" && structured.schema)
+    ? `Your JSON MUST validate EXACTLY against this JSON Schema:\n${JSON.stringify(structured.schema)}\n`
+      + `- Include every required property.\n`
+      + `- Do NOT add any property that is not defined in the schema.\n`
+      + `- Respect all declared types, enums, and nullability.`
+    : `Respond with a single valid JSON value.`;
+  let text =
+`You are a strict JSON generator. Output a SINGLE JSON value and NOTHING else.
+- The response MUST begin with { or [ and end with the matching } or ].
+- Do NOT wrap the JSON in Markdown or code fences (no \`\`\`).
+- Do NOT include any prose, explanation, heading, comment, reasoning, or XML — only the raw JSON.
+${schemaBlock}`;
+  if (attempt > 0) {
+    text = `YOUR PREVIOUS RESPONSE WAS REJECTED (${lastErr}). Output ONLY the corrected raw JSON now, with no other text.\n\n` + text;
+  }
+  return text;
+}

lib/tui/fixtures/error-401-failauth.jsonl

@@ -0,0 +1,2 @@
+{"type":"user","entrypoint":"cli","cwd":"/tmp/tui-test","sessionId":"bbbb2222-3333-4444-5555-666677778888","version":"2.1.104","message":{"role":"user","content":[{"type":"text","text":"What is 2 + 2?"}]}}
+{"type":"assistant","entrypoint":"cli","cwd":"/tmp/tui-test","sessionId":"bbbb2222-3333-4444-5555-666677778888","version":"2.1.104","message":{"role":"assistant","model":"claude-haiku-4-5-20251001","stop_reason":"end_turn","content":[{"type":"text","text":"Failed to authenticate. API Error: 401 Invalid authentication credentials"}]}}

lib/tui/fixtures/error-401.jsonl

@@ -0,0 +1,2 @@
+{"type":"user","entrypoint":"cli","cwd":"/tmp/tui-test","sessionId":"aaaa1111-2222-3333-4444-555566667777","version":"2.1.104","message":{"role":"user","content":[{"type":"text","text":"What is 2 + 2?"}]}}
+{"type":"assistant","entrypoint":"cli","cwd":"/tmp/tui-test","sessionId":"aaaa1111-2222-3333-4444-555566667777","version":"2.1.104","message":{"role":"assistant","model":"claude-haiku-4-5-20251001","stop_reason":"end_turn","content":[{"type":"text","text":"Please run /login · API Error: 401 Invalid authentication credentials"}]}}

lib/tui/fixtures/no-turn-duration.jsonl

@@ -0,0 +1,2 @@
+{"type":"user","entrypoint":"cli","cwd":"/tmp/tui-test","sessionId":"bbbb1111-2222-3333-4444-555566667777","version":"2.1.104","message":{"role":"user","content":[{"type":"text","text":"Say PONG and nothing else."}]}}
+{"type":"assistant","entrypoint":"cli","cwd":"/tmp/tui-test","sessionId":"bbbb1111-2222-3333-4444-555566667777","version":"2.1.104","message":{"role":"assistant","model":"claude-haiku-4-5-20251001","stop_reason":"end_turn","content":[{"type":"text","text":"PONG"}]}}

lib/tui/pool.mjs

@@ -0,0 +1,321 @@
+import { rmSync } from "node:fs";
+
+// TUI warm pane pool (docs/plans/2026-07-13-tui-latency backlog #3).
+//
+// WHAT IT IS: a small set of PRE-BOOTED `claude` panes, each already sitting at its
+// input bar, so a request does not pay the cold boot. Opt-in: OCP_TUI_POOL_SIZE=0
+// (default) disables it entirely and the request path is byte-for-byte today's.
+//
+// ── SINGLE-USE IS THE LOAD-BEARING RULE ─────────────────────────────────────
+// A pooled pane serves EXACTLY ONE turn and is then killed and replaced in the
+// background. Each pane carries its OWN fresh `--session-id`, fixed at boot, and the
+// turn locates its transcript by that id. So OCP's one-session-per-request model is
+// preserved: a session's transcript still holds exactly one logical exchange.
+// That is what keeps lib/tui/transcript.mjs's extractLatestAssistantText (which returns
+// the LAST text-bearing assistant entry in the whole file, not "text since the matching
+// user line") correct — see the scoping note there. A pane MUST NEVER serve a second
+// turn, and a session MUST NEVER be reset with /clear and reused: either would put two
+// exchanges in one transcript and leak the earlier turn's text into the later turn's
+// answer. Nothing here reuses a pane; keep it that way.
+//
+// ── WHY IT'S WORTH MORE THAN THE BOOT TIME ──────────────────────────────────
+// Measured on this host (n=6 through OCP, Sonnet 4.6, --effort low): the cold path
+// spends ~1.23 s reaching the input bar, but ALSO ~2.9 s inside the first turn beyond
+// what claude itself reports as the turn duration — post-input-bar init that a pane
+// which has been idle for a few seconds has already finished. A warm pane recovers both.
+//
+// ── COST (bounded, and paid whether or not a request arrives) ───────────────
+// Each warm pane is a LIVE `claude` process (plus its tmux pane) sitting idle. Peak
+// process count is (pool size) + (OCP_TUI_MAX_CONCURRENT in-flight turns) + (panes
+// currently booting as replacements). Pool size is clamped to POOL_MAX_SIZE.
+//
+// Pure + injectable (bootPane / killPane / paneHealthy / now) so test-features.mjs can
+// assert acquire / miss / refill / TTL / reaper-exemption with no tmux and no claude.
+
+// Hard cap on OCP_TUI_POOL_SIZE. Each pane is an idle claude process; 4 is already a
+// lot of resident memory on a small host (a Pi serving a family) for zero in-flight work.
+export const POOL_MAX_SIZE = 4;
+
+// A warm pane older than this is dropped on acquire rather than handed out. The periodic
+// reap tick (server.mjs) drains the pool every 15 min anyway, so this only bites when
+// that tick kept getting skipped because the TUI path was never idle. Guards against
+// handing out a pane whose `claude` has been sitting so long it may have drifted
+// (auto-compaction prompts, an idle-disconnect banner, an expired in-pane token).
+export const POOL_MAX_AGE_MS = 10 * 60 * 1000;
+
+// Clamp the operator-supplied size into [0, POOL_MAX_SIZE]. A garbage value disables the
+// pool rather than guessing — an unparseable size must never silently boot 4 processes.
+export function resolvePoolSize(raw) {
+  const n = parseInt(raw, 10);
+  if (!Number.isFinite(n) || n <= 0) return 0;
+  return Math.min(n, POOL_MAX_SIZE);
+}
+
+export class TuiPanePool {
+  // size:       target number of warm panes (0 = disabled).
+  // maxAgeMs:   per-pane TTL (see POOL_MAX_AGE_MS).
+  // mintPane:   () => ({ sessionId, name }) — mints the identity of the NEXT pane. The POOL,
+  //             not the boot function, owns this: the tmux session springs into existence the
+  //             instant bootPane starts, so the pool must already know its NAME (see
+  //             _bootingPane below). Deriving the name from the sessionId also makes `tmux ls`
+  //             correlate to the transcript file.
+  // bootPane:   async (model, {sessionId, name}) => { name, sessionId, model, bootedAt } —
+  //             boots ONE pane under exactly that identity and resolves only once it is
+  //             input-ready; throws if it never becomes ready.
+  // killPane:   (name) => void — tmux kill-session. MUST be synchronous (see drain).
+  // paneHealthy:(name) => bool — pane still exists AND is still at its input bar.
+  constructor({ size, maxAgeMs = POOL_MAX_AGE_MS, mintPane, bootPane, killPane, paneHealthy, now = Date.now, log = () => {} }) {
+    this.size = Math.max(0, Math.min(parseInt(size, 10) || 0, POOL_MAX_SIZE));
+    // Fail fast at CONSTRUCTION, not at request time. refill() is called synchronously from
+    // the request path (runTuiTurn), so a missing collaborator would otherwise surface as a
+    // 500 on a live request instead of a loud error at boot.
+    if (this.size > 0) {
+      for (const [k, fn] of [["mintPane", mintPane], ["bootPane", bootPane], ["killPane", killPane], ["paneHealthy", paneHealthy]]) {
+        if (typeof fn !== "function") throw new TypeError(`TuiPanePool: ${k} must be a function`);
+      }
+    }
+    this.maxAgeMs = maxAgeMs;
+    this._mintPane = mintPane;
+    this._bootPane = bootPane;
+    this._killPane = killPane;
+    this._paneHealthy = paneHealthy;
+    this._now = now;
+    this._log = log;
+
+    this._panes = [];        // warm, available panes: { name, sessionId, model, bootedAt }
+    // The pane currently BOOTING, BY NAME ({sessionId, name, model}) — or null.
+    //
+    // WHY A NAME AND NOT A COUNT (this is a fixed bug, don't regress it): bootTuiPane creates
+    // the tmux session SYNCHRONOUSLY and only THEN waits up to POOL_BOOT_MS (20 s) for the
+    // input bar. So for up to 20 s there is a LIVE pooled tmux session. When the pool tracked
+    // only a count, it could not NAME that session, so:
+    //   - liveNames() could not spare it and the periodic reap sweep KILLED it (and
+    //     kill-server'd on top), leaving the pool empty with nothing scheduled and firing the
+    //     very tui_pool_boot_failed WARN operators are told to alert on; and
+    //   - drain() could not kill it, so on shutdown it ORPHANED a live authenticated `claude`
+    //     (the boot's .then that was supposed to clean up never runs — gracefulShutdown calls
+    //     process.exit in the same tick).
+    // Both are fixed by holding the identity here, before the session exists.
+    this._bootingPane = null;
+    // Generation counter. Bumped whenever an in-flight boot is CANCELLED (drain / model
+    // switch). A boot compares the generation it started under against the current one:
+    // if they differ, its pane was already killed by us and its settle is inert — in
+    // particular a rejection is a CANCELLATION, not an operator-visible boot failure.
+    this._gen = 0;
+    this._paused = false;    // true while drained; refill() is a no-op until resume()
+    this.warmModel = null;   // the model the pool currently warms — learned from traffic (see acquire)
+
+    this.hits = 0;           // requests served by a warm pane
+    this.misses = 0;         // requests that fell back to the cold path
+    this.boots = 0;          // panes successfully pre-booted
+    this.bootFailures = 0;   // pre-boots that genuinely never reached the input bar
+    this.cancelled = 0;      // in-flight boots WE killed (drain / model switch) — not failures
+    this.dropped = 0;        // panes discarded unused (unhealthy / expired / wrong model / drained /
+                             // cancelled — a cancelled in-flight boot also lands here via _drop)
+  }
+
+  get enabled() { return this.size > 0; }
+  get warm() { return this._panes.length; }
+  get booting() { return this._bootingPane ? 1 : 0; }
+
+  // The reaper's spare set: the EXACT names of every pane the pool currently owns and has NOT
+  // handed out — the warm ones AND the one currently booting (whose tmux session is already
+  // live; see _bootingPane). See the POOL/REAPER INVARIANT in lib/tui/session.mjs.
+  // Fail-safe by construction: a pane leaves this set the instant it is acquired, dropped, or
+  // cancelled, and if the pool is empty (or the process restarted) the set is empty — so an
+  // orphaned pooled pane looks exactly like any other stale session and IS reaped.
+  liveNames() {
+    const names = new Set(this._panes.map((p) => p.name));
+    if (this._bootingPane) names.add(this._bootingPane.name);
+    return names;
+  }
+
+  // Take a warm pane for `model`, or null (caller must fall back to the cold path — a MISS
+  // is always safe, never an error). Synchronous: paneHealthy is a cheap tmux capture.
+  //
+  // The pool warms the MOST RECENTLY REQUESTED model (`warmModel`). There is no boot-time
+  // pre-warm and no configured model: OCP cannot know which model the next caller wants, and
+  // pre-booting a process for a model nobody asks for is pure waste. Consequence, stated
+  // plainly: the FIRST request after start (and the first after a model switch) is always a
+  // MISS. The pool pays off for the steady repeat traffic it exists to serve.
+  acquire(model) {
+    if (!this.enabled) return null;
+
+    // Retarget on a model switch: --model is fixed at spawn, so panes for another model are
+    // useless. Drop them now (they are replaced by the next refill) rather than holding
+    // processes for a model that is no longer being asked for. This includes any pane
+    // currently BOOTING for the old model — its tmux session already exists, so leaving it to
+    // die on resolve would both hold a useless process and block the next refill (one boot at
+    // a time) for up to POOL_BOOT_MS.
+    if (model !== this.warmModel) {
+      for (const p of this._panes) { this._drop(p, "model_switch"); }
+      this._panes = [];
+      this._cancelBooting("model_switch");
+      this.warmModel = model;
+    }
+
+    while (this._panes.length) {
+      const p = this._panes.shift();
+      if (this._now() - p.bootedAt > this.maxAgeMs) { this._drop(p, "expired"); continue; }
+      if (!this._paneHealthy(p.name)) { this._drop(p, "unhealthy"); continue; }
+      this.hits++;
+      return p; // caller OWNS it now: it is out of the registry (so out of the spare set),
+                // and the caller's finally MUST kill it. Single-use — never returned here.
+    }
+    this.misses++;
+    return null;
+  }
+
+  // Bring the pool back up to `size` warm panes for `warmModel`. Fire-and-forget: never
+  // awaited on the request path and never throws into it.
+  //
+  // SLOT ACCOUNTING: a refill boot deliberately does NOT take a TuiSemaphore slot. Those
+  // slots bound concurrent *turns* (each up to the 120 s wallclock) and belong to real
+  // requests; charging a background pre-boot against them would let the pool starve the
+  // traffic it exists to speed up. It cannot leak a slot either, because it never holds one.
+  //
+  // SERIALIZED, ONE BOOT AT A TIME (and re-kicked on success until the pool is at target).
+  // An earlier version launched all `want` boots at once; live at size=2 that put two cold
+  // `claude` boots plus an in-flight turn on the CPU together, and a refill overran even the
+  // generous pool readiness cap (tui_pool_boot_failed). Booting sequentially keeps each boot
+  // near its uncontended ~1.2 s, bounds the CPU burst the pool can cause, and still has the
+  // replacement pane warm long before the next request arrives.
+  //
+  // A genuinely FAILED boot deliberately does NOT re-kick the chain — that is the backoff. A
+  // persistently failing boot (bad claude binary, no auth) would otherwise spin, respawning
+  // forever. The next natural trigger (the following request's refill, or the reap tick's
+  // resume) retries it. A CANCELLED boot is different: we killed it on purpose, nothing is
+  // wrong, and resume() is expected to start a fresh one immediately.
+  refill() {
+    if (!this.enabled || this._paused || !this.warmModel) return;
+    if (this._bootingPane) return;                       // one boot in flight at a time
+    if (this._panes.length >= this.size) return;         // already at target
+
+    const model = this.warmModel;
+    const gen = this._gen;
+    // Mint the identity BEFORE booting: bootPane creates the tmux session synchronously, so
+    // the pool must be able to name (and therefore spare, and kill) it from this moment on.
+    const ident = this._mintPane();
+    this._bootingPane = { ...ident, model };
+    let enlisted = false;
+    Promise.resolve()
+      .then(() => this._bootPane(model, ident))
+      .then((pane) => {
+        // The world may have moved while we booted. If our generation was cancelled, kill the
+        // pane here rather than ASSUMING _cancelBooting already did.
+        //
+        // Why not just `return`: _cancelBooting kills by name, but the tmux session only EXISTS
+        // once _bootPane has actually run — and _bootPane is queued on a microtask (above). A
+        // caller that does refill() and then drain() in the SAME synchronous block would have
+        // _cancelBooting find nothing to kill (a no-op), bump the generation, and then this
+        // microtask would create the session, boot it fine, and — under a bare `return` — walk
+        // away from a LIVE authenticated `claude` that nothing owns. That is M1b in a new costume.
+        // No current call site does that, so this is defense-in-depth, not a live bug — but ADR
+        // 0008 and the reap-tick comment in server.mjs both explicitly contemplate a boot-time
+        // pre-warm, which is exactly the shape that would reach it.
+        //
+        // Killing an already-dead session is a harmless no-op (_drop swallows it), so this is
+        // idempotent whether or not _cancelBooting got there first.
+        if (gen !== this._gen) { this._drop(pane, "cancelled_late"); return; }
+        // Otherwise: still possible the pool filled or retargeted without a cancellation.
+        if (this._paused || model !== this.warmModel || this._panes.length >= this.size) {
+          this._drop(pane, "stale_boot");
+          return;
+        }
+        this._panes.push(pane);
+        this.boots++;
+        enlisted = true;
+      })
+      .catch((e) => {
+        // A rejection from a CANCELLED generation is not a fault: it is almost always
+        // "tui_pane_not_ready", thrown because WE killed the pane out from under the boot.
+        // Counting it as a bootFailure would fire the exact WARN operators are told to alert
+        // on, for a completely healthy drain. Stay silent — _cancelBooting already counted
+        // this as a cancellation, so do NOT count it again here.
+        if (gen !== this._gen) return;
+        this.bootFailures++;
+        this._log("warn", "tui_pool_boot_failed", { model, error: e && e.message });
+      })
+      .finally(() => {
+        // ONLY the current generation's boot owns the booting slot. A stale settle must not
+        // clear a slot that a newer boot (started by resume()) already holds.
+        if (gen === this._gen) this._bootingPane = null;
+        if (enlisted) this.refill(); // continue toward target, still one at a time
+      });
+  }
+
+  // Kill the in-flight boot's pane, SYNCHRONOUSLY, and invalidate its generation. Returns 1
+  // if there was one, else 0. The tmux session already exists (bootPane created it before it
+  // started waiting for readiness), so this is a real kill, not a cancellation flag.
+  _cancelBooting(reason) {
+    if (!this._bootingPane) return 0;
+    this._gen++;                              // the in-flight boot's settle is now inert
+    this._drop(this._bootingPane, reason);    // synchronous kill-session
+    this._bootingPane = null;
+    this.cancelled++;
+    return 1;
+  }
+
+  // Kill every pane the pool owns — warm AND currently booting — and stop refilling. Returns
+  // how many were killed.
+  //
+  // Called (a) before the periodic reap sweep — reapStaleTuiSessions can only reap defunct
+  // `claude` zombies via kill-server, and kill-server is suppressed while any live pooled pane
+  // exists (including a booting one), so without this drain the pool would permanently disable
+  // zombie reaping; and (b) on graceful shutdown, so no pane outlives the process as an orphan.
+  //
+  // EVERY KILL HERE IS SYNCHRONOUS, and that is load-bearing. It is NOT safe to leave the
+  // booting pane to clean itself up on resolve: gracefulShutdown calls process.exit() in the
+  // same tick as this drain (TUI panes are children of the tmux SERVER, not of node, so
+  // node's activeProcesses set is empty on a TUI host and the "wait for children" path exits
+  // immediately). A .then()/.catch() scheduled here would never run, and the pane would
+  // survive as an orphaned, authenticated, idle `claude`.
+  drain() {
+    this._paused = true;
+    let n = this._panes.length;
+    for (const p of this._panes) this._drop(p, "drain");
+    this._panes = [];
+    n += this._cancelBooting("drain_booting");
+    return n;
+  }
+
+  // Undo drain() and start refilling again. Because drain() CANCELLED the in-flight boot
+  // (rather than leaving it pending), the booting slot is free and this really does start a
+  // fresh boot — the pool is never left empty with nothing scheduled.
+  resume() {
+    this._paused = false;
+    this.refill();
+  }
+
+  // /health surface (additive).
+  stats() {
+    return {
+      size: this.size,
+      warm: this._panes.length,
+      booting: this.booting,
+      model: this.warmModel,
+      hits: this.hits,
+      misses: this.misses,
+      boots: this.boots,
+      bootFailures: this.bootFailures,
+      cancelled: this.cancelled,
+      dropped: this.dropped,
+    };
+  }
+
+  _drop(pane, reason) {
+    this.dropped++;
+    try { this._killPane(pane.name); } catch { /* already gone */ }
+    // F5: every drop path (expired / unhealthy / model_switch / drain / cancelled_late /
+    // stale_boot) ends up here, and the reap tick drains the WHOLE pool on every tick — so
+    // without this, every warm pane's sink orphans in streamDir with no GC path (killPane only
+    // reaches the tmux session, never the pane's OWN files). Best-effort: pane.streamFile is
+    // undefined for a still-booting identity (the sink path is only known once bootPane
+    // resolves) and rmSync(force:true) is already a no-op on a missing file, so this never
+    // throws into the reaper regardless of which drop path got here.
+    if (pane.streamFile) {
+      try { rmSync(pane.streamFile, { force: true }); } catch { /* best-effort GC */ }
+    }
+    this._log("info", "tui_pool_pane_dropped", { name: pane.name, reason });
+  }
+}

lib/tui/semaphore.mjs

@@ -0,0 +1,192 @@
+// TUI-path concurrency limiter (audit finding C-4).
+//
+// WHY THIS EXISTS, SEPARATE FROM server.mjs's MAX_CONCURRENT:
+//   The global MAX_CONCURRENT gate lives in spawnClaudeProcess() (the -p / stream-json
+//   path). callClaudeTui() NEVER calls spawnClaudeProcess — it calls runTuiTurn(), which
+//   boots a full interactive `claude` inside a fresh tmux session. So nothing bounded the
+//   TUI path: N concurrent TUI requests spawned N simultaneous cold-boot tmux+claude
+//   processes. On a small host (a Pi 4 serving a family) a burst of ~5 is an OOM risk, and
+//   it also multiplies subscription rate-limit pressure. This is an INDEPENDENT limiter for
+//   the TUI path that mirrors MAX_CONCURRENT's intent without coupling to it (the two pools
+//   are different shapes: a stream-json spawn is cheap and fast; a TUI turn is a heavy
+//   cold-boot + up to 120s wallclock).
+//
+// QUEUE vs REJECT: we QUEUE (await a slot), mirroring the spirit of MAX_CONCURRENT's
+//   intent not to drop requests, rather than rejecting immediately. To avoid unbounded
+//   memory growth from a runaway client, the wait queue itself is bounded by maxQueue
+//   (default: a generous multiple of the concurrency limit). When the queue is full, run()
+//   rejects with a tui_queue_full error (the caller surfaces it as a 503) — a deterministic
+//   backpressure signal rather than silent OOM.
+//
+// Pure + importable so test-features.mjs can assert the bound directly (no server boot).
+
+// Thrown by acquire() when the caller-supplied AbortSignal fires before a slot was granted
+// (audit finding F2 — a client that disconnects while queued must never receive a slot; the
+// queue entry is spliced out, not just flagged, so `queued` accounting stays exact). Distinct
+// `name` lets callers (server.mjs acquireClaudeSlot) tell "client went away" apart from
+// "queue is full" without string-matching the message.
+export class SemaphoreAbortError extends Error {
+  constructor(message) { super(message); this.name = "SemaphoreAbortError"; }
+}
+
+export class TuiSemaphore {
+  // limit: max concurrent slots. maxQueue: max waiters before run() rejects with backpressure.
+  constructor(limit, { maxQueue } = {}) {
+    this.limit = Math.max(1, parseInt(limit, 10) || 1);
+    // Default queue cap: 32× the limit. Large enough that real family-burst traffic never
+    // hits it, small enough that a pathological flood can't grow the queue without bound.
+    this.maxQueue = Number.isFinite(maxQueue) ? maxQueue : this.limit * 32;
+    this._inflight = 0;
+    this._waiters = []; // FIFO queue of resolve callbacks waiting for a slot
+  }
+
+  get inflight() { return this._inflight; }
+  get queued() { return this._waiters.length; }
+
+  // Runtime-adjust the concurrency limit (audit finding F1 — a PATCH /settings maxConcurrent
+  // change must actually take effect, not just be ignored until every currently-inflight task
+  // happens to finish). Lowering the limit is handled lazily by release() (see below) — it
+  // simply stops re-granting until inflight drains under the new, lower limit. Raising the
+  // limit has immediate headroom, so we wake as many queued waiters as now fit.
+  setLimit(limit) {
+    this.limit = Math.max(1, parseInt(limit, 10) || 1);
+    while (this._inflight < this.limit && this._waiters.length > 0) {
+      const next = this._waiters.shift();
+      this._inflight++;
+      next();
+    }
+  }
+
+  // Acquire a slot. Resolves once a slot is free (immediately if under the limit, otherwise
+  // when an in-flight task releases). Rejects synchronously-ish if the wait queue is full.
+  // `signal` (optional AbortSignal, F2) lets the caller cancel a QUEUED wait — e.g. wired to
+  // a client's socket "close" event so a request that disconnects before a slot is granted
+  // is removed from the queue instead of eventually being handed a slot for a dead socket.
+  // If `signal` is already aborted, reject immediately without ever touching the queue.
+  acquire(signal) {
+    if (signal?.aborted) {
+      return Promise.reject(new SemaphoreAbortError("acquire aborted before requesting a slot"));
+    }
+    if (this._inflight < this.limit) {
+      this._inflight++;
+      return Promise.resolve();
+    }
+    if (this._waiters.length >= this.maxQueue) {
+      return Promise.reject(new Error(
+        `tui_queue_full: TUI concurrency limit (${this.limit}) reached and wait queue ` +
+        `(${this.maxQueue}) is full`));
+    }
+    return new Promise((resolve, reject) => {
+      let waiter; // the FIFO entry — captured so onAbort can find + splice exactly this one
+      const onAbort = () => {
+        const idx = this._waiters.indexOf(waiter);
+        if (idx === -1) return; // already granted a slot (shifted out by release()/setLimit) — too late to cancel
+        this._waiters.splice(idx, 1); // remove, not just flag — keeps `queued` accounting exact
+        reject(new SemaphoreAbortError("acquire aborted while queued"));
+      };
+      waiter = () => {
+        signal?.removeEventListener("abort", onAbort);
+        resolve();
+      };
+      signal?.addEventListener("abort", onAbort, { once: true });
+      this._waiters.push(waiter);
+    });
+  }
+
+  // Release a slot. Always frees the caller's own slot first, then re-grants it to the next
+  // waiter ONLY if the (post-decrement) inflight count is still under the current limit (F1
+  // fix). This is what makes a runtime-lowered limit actually bite: if the limit was lowered
+  // while over-subscribed, releases stop re-granting and inflight drains toward the new limit
+  // instead of a freed slot being handed straight back out at the old, higher occupancy.
+  release() {
+    if (this._inflight > 0) this._inflight--;
+    if (this._inflight < this.limit) {
+      const next = this._waiters.shift();
+      if (next) {
+        this._inflight++;
+        next();
+      }
+    }
+  }
+
+  // Run fn() under one slot. Releases in a finally so a throw (PR-A's honesty gates,
+  // wallclock truncation, paste-not-landed, tmux spawn failure) NEVER leaks a slot.
+  // `signal` (optional, F2) is forwarded to acquire() so a queued run() can be cancelled.
+  async run(fn, signal) {
+    await this.acquire(signal);
+    try {
+      return await fn();
+    } finally {
+      this.release();
+    }
+  }
+}
+
+// ── TUI drift observability (audit C-5) — pure helpers, importable for testing ──
+
+// Record an observed cc_entrypoint into the (mutable) tuiStats counter. Sets lastEntrypoint
+// unconditionally and increments entrypointMismatches when the spawn was supposed to be
+// subscription-pool ("cli") but the transcript reported something else (a silent drift to
+// the metered Agent SDK pool — the audit's top risk after the 6/15 billing flip).
+// Returns true iff this observation was a mismatch (so the caller can also emit a log).
+export function recordTuiEntrypoint(tuiStats, observed, expectedMode = "cli") {
+  tuiStats.lastEntrypoint = observed ?? null;
+  const mismatch = expectedMode === "cli" && observed !== "cli";
+  if (mismatch) tuiStats.entrypointMismatches++;
+  return mismatch;
+}
+
+// Build the additive /health `tui` block (ADR 0007 PR-B amendment). Pure: given the
+// config + live counters, returns the exact object embedded in /health. New fields only —
+// behaviour-preserving for existing /health consumers (grandfathered B.2 under ADR 0006).
+//
+// `pool` (optional, warm pane pool — lib/tui/pool.mjs): a TuiPanePool, or null/undefined
+// when the pool is off (the default). Reported as `pool: null` when off so the block's
+// shape stays stable, and as the pool's stats (size / warm / hits / misses / …) when on —
+// the operator's window onto both the hit rate and the standing idle-process cost.
+//
+// Streaming fields (backlog #2, OCP_TUI_STREAM) are ADDITIVE too:
+//   streamEnabled       — is real (MessageDisplay-hook) SSE streaming on for TUI turns?
+//   streamTurns         — streamed turns ATTEMPTED, counted before the truncation/auth-banner
+//                         gates run (F6) — so a turn REFUSED by those gates still shows up
+//                         here, which is exactly the turn an operator most wants visible.
+//                         Counting only turns that survived the gates would silently exclude
+//                         a turn's worst-case outcome from its own denominator.
+//   streamDeltas        — MessageDisplay hook fires OBSERVED, including held-back ones (F6) —
+//                         NOT only the ones forwarded to a client. This is what makes
+//                         streamZeroDeltaTurns meaningful: a turn can have streamDeltas
+//                         incrementing while still emitting nothing to the client (fully held
+//                         back, e.g. a short answer), which is healthy, vs. a hook that fired
+//                         zero times at all, which is not (see streamZeroDeltaTurns).
+//   streamTopUps        — turns where the delta stream was a safe PREFIX of the transcript but
+//                         not equal to it; OCP topped up from the transcript and served T.
+//                         Benign but worth watching — a persistent rate means the hook is
+//                         losing fires.
+//   streamDivergences   — turns REFUSED because emitted bytes were not a prefix of the
+//                         transcript. THE field to alert on for CORRECTNESS: it means the hook
+//                         and the transcript disagreed and OCP chose to fail rather than serve
+//                         unverifiable text.
+//   streamZeroDeltaTurns — streamed turns where the hook fired ZERO times (F7). THE field to
+//                         alert on for AVAILABILITY: streamTopUps climbing is one fire dropped
+//                         here and there (benign); this climbing means the hook is not firing
+//                         AT ALL — e.g. `--settings` silently stopped registering it (a claude
+//                         version bump), or F3's truncated-script failure mode — and every
+//                         streamed turn is quietly degrading to fully-buffered with no error.
+export function buildTuiHealthBlock({ enabled, entrypointMode, maxConcurrent, streamEnabled = false }, tuiStats, semaphore, pool = null) {
+  return {
+    enabled,
+    entrypointMode,                                  // cli | auto | off
+    lastEntrypoint: tuiStats.lastEntrypoint,         // last observed cc_entrypoint, or null
+    entrypointMismatches: tuiStats.entrypointMismatches,
+    inflight: semaphore.inflight,                    // current concurrent TUI turns
+    queued: semaphore.queued,                        // turns waiting for a slot
+    maxConcurrent,
+    pool: pool ? pool.stats() : null,                // warm pane pool, or null when disabled
+    streamEnabled,
+    streamTurns: tuiStats.streamTurns ?? 0,
+    streamDeltas: tuiStats.streamDeltas ?? 0,
+    streamTopUps: tuiStats.streamTopUps ?? 0,
+    streamDivergences: tuiStats.streamDivergences ?? 0,
+    streamZeroDeltaTurns: tuiStats.streamZeroDeltaTurns ?? 0,
+  };
+}

lib/tui/session.mjs

@@ -0,0 +1,782 @@
+// TUI-mode session driver: hosts an interactive `claude` in tmux, submits one
+// serialized prompt, awaits the transcript reader, tears down. OCP-specific.
+//
+// Authority: claude CLI v2.1.158 interactive mode (no -p / no --output-format
+// => cc_entrypoint=cli). Submission recipe validated by spikes T3/T6 on PI231.
+// See docs/superpowers/specs/2026-05-30-tui-mode-production-design.md.
+//
+// Trust handling: rather than answer the trust-folder dialog interactively (which
+// only appears on a cwd's FIRST encounter — sending a defensive "1" to an already
+// trusted cwd would inject a stray prompt turn), we PRE-TRUST the scratch cwd by
+// seeding <home>/.claude.json. Every turn then boots dialog-free and identical.
+import { spawnSync } from "node:child_process";
+import { mkdtempSync, writeFileSync, readFileSync, mkdirSync, existsSync, rmSync, statSync, renameSync, symlinkSync } from "node:fs";
+import { tmpdir } from "node:os";
+import { randomUUID } from "node:crypto";
+import { readTuiTranscript } from "./transcript.mjs";
+import { prepareStreamHook, streamFilePath, parseDeltaChunk } from "./stream.mjs";
+
+// F7 fix (audit finding, LOW): the prefix used to be a bare, host-wide constant
+// ("ocp-tui-"), so a SECOND OCP instance on the same host (e.g. a temporary
+// verification instance stood up alongside production — a real pattern used during
+// PR #144/#146 verification) would boot-reap and potentially kill-server the OTHER
+// instance's LIVE sessions: the coexistence guard below only ever spared foreign
+// PRODUCT prefixes (olp-tui-*), never a second ocp-tui-* instance on a different port.
+//
+// Fix: scope the prefix to the instance's own listen port. The port is the natural
+// stable per-instance discriminator on one host (two OCP instances cannot share a
+// port), so `ocp-tui-<port>-` uniquely namespaces this instance's sessions and makes
+// a same-host sibling OCP instance look exactly like a foreign product (olp-tui-*) to
+// the coexistence guard — its `ocp-tui-<otherPort>-*` sessions never match our own
+// prefix and are therefore never reaped/kill-server'd by us.
+//
+// LEGACY_SESSION_PREFIX / LEGACY_SESSION_NAME_RE describe the OLD bare-prefix shape
+// (pre-this-fix), retained ONLY for the boot-time legacy-zombie migration handled in
+// reapStaleTuiSessions (see comment there). No code path in this version ever CREATES
+// a legacy-shaped session name again — sessionPrefixForPort() is the only session-name
+// prefix constructor used going forward.
+export const LEGACY_SESSION_PREFIX = "ocp-tui-";
+// Exact legacy shape: LEGACY_SESSION_PREFIX + sessionId.slice(0, 8), where sessionId is
+// a randomUUID() — so the suffix is always exactly 8 lowercase hex characters with NO
+// further separator. The new port-scoped shape always inserts a "-" between the port
+// digits and the 8-hex suffix (see sessionPrefixForPort), so this regex can never match
+// a new-shape name: a new-shape suffix is `<port digits>-<8 hex>` (contains a literal
+// "-"), which `[0-9a-f]{8}$` anchored immediately after the prefix cannot satisfy.
+export const LEGACY_SESSION_NAME_RE = /^ocp-tui-[0-9a-f]{8}$/;
+
+// Build this instance's own session-name prefix, scoped by its listen port so a
+// second OCP instance on the same host (different port) is never mistaken for "ours".
+export function sessionPrefixForPort(port) {
+  return `ocp-tui-${port}-`;
+}
+
+const TMUX = process.env.OCP_TUI_TMUX_BIN || "tmux";
+
+const defaultTmux = (args, opts = {}) =>
+  spawnSync(TMUX, args, { encoding: "utf8", ...opts });
+
+// Kill ONLY our own stale sessions. Scoped to sessionPrefixForPort(port) so a co-hosted
+// OLP test instance's `olp-tui-*` sessions — AND a co-hosted second OCP instance's
+// `ocp-tui-<otherPort>-*` sessions — are never touched (F7 fix).
+//
+// Defunct-reaping (PI231 incident): the pane's `claude` process is a child of the
+// long-lived tmux SERVER daemon, NOT of the OCP node process — `tmux new-session -d`
+// returns the instant the server forks the pane, so node never becomes its parent and
+// therefore can NEVER waitpid()/reap it (a SIGKILL still needs the *parent* to reap, and
+// here that parent is the tmux server). `kill-session` destroys the session but the server
+// can leave the pane's `claude` (and any grandchildren claude spawned) as `<defunct>`
+// zombies that only the server can reap. Over many per-request spawn+teardown cycles these
+// accumulate (live evidence on PI231: 25 defunct `<claude>` over 30 days; `tmux kill-server`
+// dropped it 25→3). The only node-reachable action that ACTUALLY reaps them — rather than
+// merely re-signalling — is to stop the tmux server: when the server exits, the kernel
+// reparents its surviving children to init (PID 1), which reaps them immediately.
+//
+// `port` (required) is this instance's own listen port (server.mjs's PORT / lib/constants.mjs
+// DEFAULT_PORT resolution) — the SPOT for "which sessions are ours."
+//
+// ── POOL/REAPER INVARIANT (warm pane pool — lib/tui/pool.mjs) ───────────────────────────
+// A warm pooled pane is one of OUR OWN `ocp-tui-<port>-*` sessions that is ALIVE AND IDLE
+// BY DESIGN — and the periodic sweep runs precisely when the instance is idle, i.e. exactly
+// when the pool is full. Without an exemption the sweep would kill every warm pane on every
+// tick (and kill-server on top). The exemption is `spare`: a set of EXACT session names the
+// caller declares live. Three properties, all load-bearing:
+//
+//   1. A LIVE POOLED PANE IS NEVER REAPED — INCLUDING ONE THAT IS STILL BOOTING. It is in
+//      `spare` (the pool's live registry), so it is skipped by name. The booting case is not
+//      a footnote, it is the one that bit us: bootTuiPane creates the tmux session
+//      SYNCHRONOUSLY and only then waits up to POOL_BOOT_MS for the input bar, so a pooled
+//      session can be live for ~20 s before its boot resolves. The pool therefore mints the
+//      pane's NAME up front and holds it in `_bootingPane`, so liveNames() can name — and
+//      spare — a session whose boot has not finished. (An earlier version tracked only a
+//      COUNT of in-flight boots; the sweep could not name that session and killed it.)
+//   2. A LEAKED/ORPHANED POOLED PANE IS STILL REAPED. Membership is by EXACT NAME from a
+//      live in-memory registry — NOT by "looks pooled" (name shape). A pane the pool no
+//      longer owns (handed out, dropped, cancelled, or left behind by a previous process
+//      generation — whose registry died with it) is absent from `spare` and is killed like
+//      any other stale session. Fail-safe: forgetting to pass `spare` reaps MORE, never less.
+//   3. KILL-SERVER NEVER KILLS A LIVE POOL PANE. A spared session suppresses kill-server
+//      exactly as a foreign session does (it is a live child of the tmux server). The
+//      consequence — that a permanently-full pool would permanently disable the defunct-
+//      zombie reaping that ONLY kill-server can do — is resolved in server.mjs by DRAINING
+//      the pool immediately before the sweep, so `spare` is empty on the normal tick and
+//      kill-server still fires. `spare` is the belt-and-braces: a reap call site that
+//      forgets to drain still cannot kill a live pane.
+//
+// `spare` (default: none) — iterable of session names, or a Set. Ignored when the pool is off.
+//
+// `includeLegacy` (default false): when true, sessions matching the exact OLD bare-prefix
+// shape (LEGACY_SESSION_NAME_RE) are ALSO treated as ours for kill-session purposes. This is
+// the boot-time legacy migration: an operator upgrading past this fix could otherwise be left
+// with orphaned bare-prefix zombie sessions from the PREVIOUS (pre-fix) process generation of
+// this SAME instance, since no live instance of the new version ever creates that shape again
+// — a legacy-shaped session found at boot is therefore presumed to be this instance's own
+// leftover, not a stranger's. Passed true ONLY from the one-time boot-reap call site in
+// server.mjs; the periodic idle-reap sweep does NOT set it, so a lingering legacy session
+// during steady-state is conservatively treated as foreign (correctly blocking kill-server)
+// rather than assumed to be ours on every 15-minute tick. Residual (accepted, documented):
+// if a genuinely-still-running PRE-FIX OCP instance is coexisting on the same host at the
+// exact moment a new instance boots, its live legacy-shaped session could be reaped — the
+// same class of residual risk the audit finding itself accepts ("no live instance of the new
+// version creates them"); this PR does not regress that scenario, it only removes the far
+// more common same-version collision (the actual F7 finding).
+export function reapStaleTuiSessions({ tmux = defaultTmux, port, includeLegacy = false, spare = null } = {}) {
+  const r = tmux(["list-sessions", "-F", "#{session_name}"]);
+  if (!r || r.status !== 0) return 0; // no tmux server / no sessions
+  const names = String(r.stdout || "").split("\n").map((s) => s.trim()).filter(Boolean);
+  const ownPrefix = sessionPrefixForPort(port);
+  const spared = spare instanceof Set ? spare : new Set(spare || []);
+  let killed = 0;
+  let othersRemain = false;
+  let sparedLive = 0;
+  for (const name of names) {
+    // Property 1+2: exemption is by EXACT NAME from the pool's live registry. A pooled-
+    // LOOKING name that is not in the registry is an orphan and falls through to the
+    // normal kill path below.
+    if (spared.has(name)) { sparedLive++; continue; }
+    const isOwn = name.startsWith(ownPrefix);
+    const isLegacyOwn = includeLegacy && LEGACY_SESSION_NAME_RE.test(name);
+    if (isOwn || isLegacyOwn) {
+      tmux(["kill-session", "-t", name]);
+      killed++;
+    } else {
+      othersRemain = true; // a session we do NOT own (olp-tui-*, a sibling ocp-tui-<otherPort>-*,
+                            // or — outside includeLegacy — a legacy-shaped name) — never kill-server
+    }
+  }
+  // Reap defunct `claude` zombies: safe ONLY when the server is now ours-only/empty.
+  // kill-server is what actually reaps (server exit reparents survivors to init); a
+  // per-session kill cannot, since node is not the zombies' parent.
+  //
+  // Property 3: a SPARED session is a live child of this tmux server, so kill-server would
+  // kill it — it therefore suppresses kill-server exactly as a foreign session does. On the
+  // normal sweep the pool is drained first, so sparedLive is 0 and kill-server still fires.
+  if (!othersRemain && sparedLive === 0) {
+    tmux(["kill-server"]);
+  }
+  return killed;
+}
+
+// ── Task 5: runTuiTurn ───────────────────────────────────────────────────
+
+// Boot + paste-settle timing. Conservative defaults validated on PI231; env-tunable.
+const BOOT_MS         = parseInt(process.env.OCP_TUI_BOOT_MS  || "4000", 10);   // max wait for input-ready
+// Readiness cap for a POOL pre-boot. Deliberately far more generous than BOOT_MS: BOOT_MS is
+// tight because a client is blocked on it, whereas a warm-pane boot happens in the background
+// with nobody waiting. Observed live at size=2: a refill booting alongside an in-flight turn
+// exceeded 4000 ms and was discarded (tui_pool_boot_failed), quietly costing hit rate for a
+// pane that was merely slow, not broken. Scales with OCP_TUI_BOOT_MS if an operator raises it.
+export const POOL_BOOT_MS = BOOT_MS * 5;
+const READY_POLL_MS   = parseInt(process.env.OCP_TUI_READY_POLL_MS || "400", 10); // readiness / paste-verify poll interval
+const PASTE_VERIFY_MS = parseInt(process.env.OCP_TUI_PASTE_VERIFY_MS || "5000", 10); // max wait for pasted prompt to render
+// Hook-sink drain interval when streaming. 100ms: the hook fires at BLOCK granularity
+// (~5-7 fires per answer, seconds apart), so a finer poll buys nothing and a coarser one
+// would add visible lag to the first delta. Cheap — one readFileSync of a small file.
+const STREAM_POLL_MS  = parseInt(process.env.OCP_TUI_STREAM_POLL_MS || "100", 10);
+
+const sleep = (ms) => new Promise((r) => setTimeout(r, ms));
+
+// Capture the visible tmux pane as plain text (for readiness / paste verification).
+function tuiCapturePane(tmux, tmuxName) {
+  const r = tmux(["capture-pane", "-p", "-t", tmuxName]);
+  return (r && typeof r.stdout === "string") ? r.stdout : "";
+}
+
+// True once claude's input bar is rendered and ready for keystrokes.
+function tuiInputReady(pane) {
+  return /\? for shortcuts/.test(pane);
+}
+
+// True once the pasted prompt has POSITIVELY landed in the input box. We only trust
+// affirmative signals — NOT "the placeholder is gone", which is unreliable (claude's
+// placeholder uses a curly quote `"`, randomized example text, and renders the big paste
+// a beat after paste-buffer returns; a "placeholder-gone" heuristic false-positived on the
+// still-empty box and made us submit Enter into nothing → issue #130 hang). Landed iff:
+//   (a) the bracketed-paste indicator "[Pasted text" is present (large/multi-line paste), OR
+//   (b) the prompt's own leading text appears in the pane (short/literal paste).
+function tuiPromptLanded(pane, prompt) {
+  const flatPane = pane.replace(/\s+/g, " ");
+  if (flatPane.includes("[Pasted text")) return true;
+  const firstLine = String(prompt).split("\n").map(s => s.trim()).find(Boolean) || "";
+  const needle = firstLine.replace(/\s+/g, " ").slice(0, 24);
+  // C-4/#133: threshold lowered 3 → 2. A prompt whose first non-blank line is 1–2
+  // chars ("hi", "ok") previously NEVER matched (needle.length >= 3) and never
+  // surfaced "[Pasted text", so EVERY short prompt 5s-failed with tui_paste_not_landed
+  // (live-reproduced: "hi"). The input box starts EMPTY (the curly-quote placeholder
+  // is excluded by the affirmative-signal design above), so a >=2-char needle present
+  // in the pane is the pasted prompt, not placeholder noise — false-positive risk is
+  // low. We keep >=2 rather than >=1 because a single visible char is more likely to
+  // collide with incidental glyphs in claude's chrome (borders, the "❯" prompt mark);
+  // 2 chars is the floor that lands real prompts while staying conservative.
+  return needle.length >= 2 && flatPane.includes(needle);
+}
+
+async function pollUntil(fn, { timeoutMs, intervalMs }) {
+  const deadline = Date.now() + timeoutMs;
+  while (Date.now() < deadline) {
+    try { if (fn()) return true; } catch { /* ignore, keep polling */ }
+    await sleep(intervalMs);
+  }
+  return false;
+}
+
+// Single-quote escaper for sh -c arguments.
+function shq(s) {
+  return `'${String(s).replace(/'/g, "'\\''")}'`;
+}
+
+// Pre-trust the scratch cwd by seeding the trust record in <home>/.claude.json so
+// the trust-folder dialog never appears. Verified-live trust shape:
+//   projects["<cwd>"] = { hasTrustDialogAccepted: true, allowedTools: [], ... }
+// Idempotent + best-effort: a missing/unreadable .claude.json must not abort a
+// turn (a fresh cwd would then show the dialog once; the boot wait tolerates it).
+// Must run BEFORE the session boots so claude reads the trusted record at startup.
+export function ensureTuiCwdTrusted(home, cwd) {
+  if (!home || !cwd) return;
+  const path = `${home}/.claude.json`;
+  let j, mode;
+  try {
+    j = JSON.parse(readFileSync(path, "utf8"));
+    mode = statSync(path).mode & 0o777;
+  } catch { return; }
+  j.projects = j.projects || {};
+  const entry = j.projects[cwd] || {};
+  if (entry.hasTrustDialogAccepted === true) return; // already trusted, no rewrite
+  entry.hasTrustDialogAccepted = true;
+  if (!Array.isArray(entry.allowedTools)) entry.allowedTools = [];
+  j.projects[cwd] = entry;
+  // Atomic write (temp + rename on the same fs), preserving mode, so a crash
+  // mid-write can never truncate the user's real ~/.claude.json. We seed ONLY the
+  // per-project trust flag — NOT bypassPermissionsModeAccepted: the driver never
+  // passes --dangerously-skip-permissions, so the bypass dialog cannot appear, and
+  // onboarding completion is an A-path precondition (the host already runs claude).
+  // NOTE: when the A-path moves to a dedicated scratch HOME (task #26), this writes
+  // a file we fully own, removing the real-config-mutation concern entirely.
+  try {
+    const tmp = `${path}.ocp-tui.${process.pid}.tmp`;
+    writeFileSync(tmp, JSON.stringify(j, null, 2), { mode });
+    renameSync(tmp, path);
+  } catch { /* best effort */ }
+}
+
+// Resolve the HOME the TUI `claude` runs under. Three intents, decided by the env
+// token + an explicit OCP_TUI_HOME override:
+//
+//   - ENV-TOKEN MODE (default when CLAUDE_CODE_OAUTH_TOKEN is set AND OCP_TUI_HOME is
+//     unset): a CREDENTIAL-FREE scratch home at `<realHome>/.ocp-tui/home`. There is
+//     deliberately NO .credentials.json (no symlink, no copy), so the only credential
+//     claude can find is the long-lived env token (passed by buildTuiCmd). This is what
+//     actually FORCES env-token auth — see the prepareTuiHome comment for why passing
+//     the token alone is insufficient.
+//   - EXPLICIT OVERRIDE: whatever OCP_TUI_HOME names (back-compat; an operator who set it
+//     keeps exactly that home).
+//   - REAL-HOME (default when the env token is unset): the operator's real home, shared
+//     credentials.json — byte-for-byte the pre-fix behaviour for credentials.json hosts.
+//
+// Pure + deterministic so server.mjs and the tests share one decision. `configuredHome`
+// is the raw OCP_TUI_HOME value (undefined/empty => unset).
+export const DEFAULT_TUI_SCRATCH_HOME = (realHome) => `${realHome}/.ocp-tui/home`;
+export function resolveTuiHome({ realHome, configuredHome, envTokenSet }) {
+  if (configuredHome) return configuredHome;            // explicit override wins (back-compat)
+  if (envTokenSet) return DEFAULT_TUI_SCRATCH_HOME(realHome); // credential-free scratch
+  return realHome;                                       // legacy real-home default
+}
+
+// Prepare the HOME claude runs under. Three modes:
+//   - real-home (tuiHome === realHome OR falsy): no isolation; just trust the cwd
+//     in the real ~/.claude.json. The legacy default when no env token is set.
+//   - ENV-TOKEN scratch-home (envTokenMode === true): a dedicated HOME with a seeded
+//     .claude.json (onboarded + trusts only the scratch cwd) and its own projects/ dir,
+//     and DELIBERATELY NO .credentials.json (no symlink, no copy). claude then has no
+//     credentials file to read, so it authenticates via CLAUDE_CODE_OAUTH_TOKEN (passed
+//     by buildTuiCmd) — which is authoritative precisely because nothing shadows it.
+//   - legacy scratch-home (envTokenMode falsy, tuiHome !== realHome): the historical
+//     mode that SYMLINKS the real .credentials.json. Retained only for an operator who
+//     explicitly set OCP_TUI_HOME without an env token; see the caveat below.
+//
+// WHY ENV-TOKEN MODE IS THE FIX (proven live on PI231, claude 2.1.104):
+//   env token passed + a broken ~/.claude/.credentials.json present → 401.
+//   env token passed + credentials.json moved aside              → real answer.
+// Interactive `claude` PREFERS .credentials.json over the env var (unlike `-p`, where the
+// env token wins), so a stale/corrupt credentials.json SHADOWS the env token. Passing the
+// token is necessary but insufficient; the TUI claude must run in a HOME with NO
+// credentials.json so the env token is the only credential. This ALSO ends the refresh-
+// corruption incident at the root: with no credentials file, claude never runs the token-
+// refresh path, so the single-use refresh token can never be rotated (and corrupted) by the
+// spawn+kill cycle. (This RESOLVES — not reintroduces — the ADR 0007 scratch-home concern:
+// the old caveat was about a SYMLINKED credentials.json being forked on refresh; here there
+// is no credentials file to fork and no refresh ever happens.)
+//
+// ⚠️ LEGACY SCRATCH-HOME CAVEAT (envTokenMode falsy, symlink path): claude rewrites
+// .credentials.json on token refresh, REPLACING the symlink with a regular-file copy → the
+// scratch home FORKS the OAuth credentials and a refresh can invalidate the real-home token.
+// That path is therefore safe only with a DEDICATED OAuth or for ephemeral use. The env-token
+// mode above avoids this entirely.
+//
+// Idempotent + best-effort: any failure degrades toward the dialog/cap, never corrupts.
+// Run BEFORE the session boots.
+export function prepareTuiHome(realHome, tuiHome, cwd, { envTokenMode = false } = {}) {
+  if (!tuiHome || tuiHome === realHome) { ensureTuiCwdTrusted(realHome, cwd); return; }
+  try {
+    const claudeDir = `${tuiHome}/.claude`;
+    mkdirSync(`${claudeDir}/projects`, { recursive: true });
+    if (!envTokenMode) {
+      // Legacy mode ONLY: symlink the real credentials (never copy the token); refresh if
+      // missing. Env-token mode deliberately skips this — no credentials file at all.
+      const link = `${claudeDir}/.credentials.json`;
+      if (!existsSync(link)) {
+        try { symlinkSync(`${realHome}/.claude/.credentials.json`, link); } catch { /* best effort */ }
+      }
+    }
+    // Seed .claude.json ONCE (if absent): onboarded + trust ONLY the scratch cwd.
+    // In env-token mode start from a MINIMAL config (do NOT copy the real ~/.claude.json —
+    // a credential-isolated home should not inherit the operator's account/config state);
+    // in legacy mode carry the onboarded real config minus the user's project history.
+    const seedPath = `${tuiHome}/.claude.json`;
+    if (!existsSync(seedPath)) {
+      let base = {};
+      if (!envTokenMode) {
+        try { base = JSON.parse(readFileSync(`${realHome}/.claude.json`, "utf8")); } catch { /* fresh */ }
+      }
+      base.hasCompletedOnboarding = true;
+      base.projects = { [cwd]: { hasTrustDialogAccepted: true, allowedTools: [] } };
+      writeFileSync(seedPath, JSON.stringify(base, null, 2), { mode: 0o600 });
+    }
+  } catch { /* best effort */ }
+  // Ensure the cwd is trusted in the scratch config (idempotent; atomic).
+  ensureTuiCwdTrusted(tuiHome, cwd);
+}
+
+// Build interactive claude argv: NO -p, NO --output-format (=> cc_entrypoint=cli).
+// MCP hard-disabled: --strict-mcp-config (no --mcp-config) is the only mechanism
+// that stops account-attached managed MCP from connecting (spec §5.2 / T6),
+// belt-and-braces with --disallowedTools "mcp__*".
+// A-PATH ONLY: built-in tools are left enabled (acceptable single-user). Deployment B
+// (guest keys) MUST additionally pass --tools "" per spec §5.2(2) as the credential
+// wall before this argv is reachable for owner_tier=guest — guard that in PR-3 wiring.
+//
+// `stream` (optional, OCP_TUI_STREAM): { file, settings } — when present, the pane gets
+//   (a) OCP_TUI_STREAM_FILE in its env  — read by the static MessageDisplay hook script to
+//       decide WHERE to append this pane's deltas. Delivered as env (not baked into the
+//       settings file) so the settings file stays STATIC and a pre-booted warm pane works.
+//       Verified live: a claude hook inherits the pane's environment.
+//   (b) --settings <file>               — registers the MessageDisplay hook.
+//   VERIFIED LIVE (claude 2.1.207, this host) before shipping, because both were spawn-level
+//   risks:
+//     - the startup banner is UNCHANGED with --settings: "Sonnet 4.6 with low effort ·
+//       Claude Max" (subscription pool). --settings is NOT a --bare-class flag — it does not
+//       silently drop the subscription pool. Transcript entrypoint stayed "cli".
+//     - --settings MERGES into the settings hierarchy, it does NOT clobber <HOME>/.claude/
+//       settings.json: with --settings passed, the user-level settings.json's `env` block was
+//       still applied to the hook's environment. So the isolated-HOME settings story the TUI
+//       already relies on (permissions / additionalDirectories — see prepareTuiHome and the
+//       OCP_TUI_FULL_TOOLS note above) survives intact.
+//   When absent, the argv is byte-for-byte the pre-streaming argv.
+export function buildTuiCmd(claudeBin, model, sessionId, ehome, entrypointMode, stream = null) {
+  // Deliver claude's env via an `env` prefix on the PANE COMMAND — tmux does NOT forward the
+  // spawning process's environment to the pane, and `new-session -e` needs tmux ≥3.2 (the cloud
+  // host runs 2.7), so this is the only portable, reliable mechanism (verified live 2026-06-01:
+  // passing {env} to spawnSync left the pane with only HOME). DISABLE_AUTOUPDATER pins the version
+  // (no "What's new" splash that delayed input-readiness); CLAUDE_CODE_ENTRYPOINT labels the
+  // billing pool (set below per entrypointMode).
+  //
+  // CLAUDE_CODE_DISABLE_CLAUDE_MDS + DISABLE_AUTO_MEMORY: OCP is a PROXY, not a Claude Code
+  // session. The proxied client (OpenClaw / an IDE) owns its own context and memory; the HOST's
+  // CLAUDE.md and auto-memory must NEVER leak into the agent OCP runs on the user's behalf.
+  // Without these, claude loads the host's project/user CLAUDE.md + memory into every proxied
+  // turn — verified live 2026-06-02: a cwd CLAUDE.md ("end every reply with QUACKMARKER_42") was
+  // obeyed by the proxied turn until these flags were set, after which it was not. Unconditional
+  // by design (not gated): proxy purity is not an opt-in. Harmless on hosts with no CLAUDE.md
+  // (the common case — they suppress nothing). Mirrors the -p path's CLAUDE_NO_CONTEXT vars.
+  const sets = [
+    `HOME=${shq(ehome)}`,
+    "DISABLE_AUTOUPDATER=1",
+    "CLAUDE_CODE_DISABLE_OFFICIAL_MARKETPLACE_AUTOINSTALL=1",
+    "CLAUDE_CODE_DISABLE_CLAUDE_MDS=1",
+    "CLAUDE_CODE_DISABLE_AUTO_MEMORY=1",
+  ];
+  // CLAUDE_CODE_OAUTH_TOKEN: tmux does NOT forward the parent process's env to the pane (the
+  // same reason the whole env is delivered as an `env` prefix above — verified live 2026-06-01),
+  // so the token MUST be set explicitly here or the spawned `claude` never sees it. Without it,
+  // the TUI claude falls back to authenticating via <HOME>/.claude/.credentials.json, whose
+  // single-use refresh token gets corrupted by the per-request spawn + `kill-session` teardown
+  // racing claude's token-rotation write (the PI231 incident: refresh token ended up an empty
+  // string → permanent 401 "Please run /login", re-login re-corrupted on the next spawn). With
+  // the long-lived OAuth token in env, claude authenticates via the token and never touches the
+  // credentials.json refresh path — matching how the stable oracle / Mac-mini hosts already run.
+  //
+  // SECURITY: the token appears in the pane command (ps-visible). This is acceptable for the
+  // single-user A-path — it mirrors the existing plaintext-token practice (server.mjs reads the
+  // same CLAUDE_CODE_OAUTH_TOKEN env at getOAuthCredentials()), and the multi-user B-path is
+  // already refused at boot (TUI + AUTH_MODE=multi is a hard FATAL). Read from process.env here,
+  // consistent with how buildTuiCmd already reads OCP_TUI_FULL_TOOLS / CLAUDE_ALLOWED_TOOLS below.
+  //
+  // When the env is unset (e.g. a host that intentionally relies on credentials.json), no token
+  // is added — behaviour is byte-for-byte unchanged from before this fix.
+  if (process.env.CLAUDE_CODE_OAUTH_TOKEN) {
+    sets.push(`CLAUDE_CODE_OAUTH_TOKEN=${shq(process.env.CLAUDE_CODE_OAUTH_TOKEN)}`);
+  }
+  // Streaming sink: the pane's own per-session delta file (see the `stream` note above).
+  if (stream && stream.file) sets.push(`OCP_TUI_STREAM_FILE=${shq(stream.file)}`);
+  const unset = ["CLAUDECODE", "ANTHROPIC_API_KEY", "ANTHROPIC_BASE_URL", "ANTHROPIC_AUTH_TOKEN"];
+  if (entrypointMode === "cli") sets.push("CLAUDE_CODE_ENTRYPOINT=cli");
+  else if (entrypointMode === "auto") unset.push("CLAUDE_CODE_ENTRYPOINT"); // let claude self-classify via TTY
+  const envPrefix = ["env", ...unset.map((u) => `-u ${u}`), ...sets].join(" ");
+
+  // Tool surface.
+  //   DEFAULT (safe): hard-disable MCP (--strict-mcp-config + --disallowedTools mcp__*);
+  //     built-in tools stay on, acceptable for single-user A-path.
+  //   OCP_TUI_FULL_TOOLS=1: grant the SAME tool surface as the -p A-path
+  //     (--allowedTools [+ --mcp-config]), so a SINGLE-USER / trusted TUI deployment can
+  //     run a tool-using agent (e.g. an OpenClaw assistant that needs Bash/Read/Write/MCP)
+  //     on the subscription pool. ALWAYS uses --allowedTools (CLAUDE_SKIP_PERMISSIONS /
+  //     --dangerously-skip-permissions is intentionally removed: claude v2.1.x shows an
+  //     interactive bypass-acceptance screen in headless tmux that nothing can answer →
+  //     the turn hangs until the wallclock cap, bricks the pane; not recoverable without a
+  //     human at a keyboard). Use scratch-home settings.json additionalDirectories instead.
+  let toolArgs;
+  if (process.env.OCP_TUI_FULL_TOOLS === "1") {
+    toolArgs = [];
+    const allowed = (process.env.CLAUDE_ALLOWED_TOOLS ||
+      "Bash,Read,Write,Edit,Glob,Grep,WebSearch,WebFetch,Agent")
+      .split(",").map((s) => s.trim()).filter(Boolean);
+    // shq EACH token: buildTuiCmd returns a SHELL STRING (run by tmux via sh -c), unlike
+    // buildCliArgs which returns an argv array to spawn(). claude accepts scoped specifiers
+    // like "Bash(npm run test:*)" / "Read(~/**)" whose ( ) * ~ would break/inject the shell
+    // command if pasted bare. (operator-self-injection only — guests can't reach TUI.)
+    if (allowed.length) toolArgs.push("--allowedTools", ...allowed.map(shq));
+    if (process.env.CLAUDE_MCP_CONFIG) toolArgs.push("--mcp-config", shq(process.env.CLAUDE_MCP_CONFIG));
+  } else {
+    toolArgs = ["--strict-mcp-config", "--disallowedTools", shq("mcp__*")];
+  }
+
+  // Effort: pass --effort EXPLICITLY. Without it, the pane's claude inherits a
+  // HOME-dependent effortLevel — real-home mode inherits the operator's
+  // ~/.claude/settings.json (whatever they set for their own interactive use),
+  // env-token scratch mode inherits claude's built-in default (prepareTuiHome never
+  // writes effortLevel) — so latency silently depends on which HOME mode
+  // resolveTuiHome() picked AND on an unrelated operator setting. Pinning it here
+  // removes both. Measured (docs/plans/2026-07-13-tui-latency): explicit low cuts
+  // direct-spawn TTFT p50 10.35s → 6.17s (−40%) and collapses the spread ~15×;
+  // banner-verified to stay on the subscription pool (`· Claude Max`).
+  // OCP_TUI_EFFORT=inherit restores the pre-flag argv byte-for-byte (no --effort).
+  // An unknown value falls back to the default rather than reaching claude's argv:
+  // a typo'd --effort value must not risk a spawn-time usage error in the pane.
+  const EFFORT_LEVELS = ["low", "medium", "high", "xhigh", "max"]; // claude 2.1.207 --help
+  const effortRaw = (process.env.OCP_TUI_EFFORT || "low").trim().toLowerCase();
+  let effortArgs;
+  if (effortRaw === "inherit") {
+    effortArgs = [];
+  } else if (EFFORT_LEVELS.includes(effortRaw)) {
+    effortArgs = ["--effort", effortRaw];
+  } else {
+    console.error(`[tui] invalid OCP_TUI_EFFORT=${JSON.stringify(process.env.OCP_TUI_EFFORT)}; using "low" (valid: ${EFFORT_LEVELS.join("|")}, or "inherit" to omit the flag)`);
+    effortArgs = ["--effort", "low"];
+  }
+
+  // --settings registers the MessageDisplay hook. Omitted entirely when streaming is off,
+  // so the OFF argv is byte-for-byte the pre-streaming argv.
+  const settingsArgs = stream && stream.settings ? ["--settings", shq(stream.settings)] : [];
+
+  return [
+    envPrefix,
+    shq(claudeBin),
+    "--model",          shq(model),
+    "--session-id",     sessionId,
+    ...toolArgs,
+    ...effortArgs,
+    ...settingsArgs,
+  ].join(" ");
+}
+
+// Is a pane alive AND still sitting at its input bar? Used by the warm pool to decide,
+// at hand-out time, whether a pre-booted pane is still usable (a dead/degraded pane must
+// become a MISS → cold path, never a hung turn). capture-pane exits non-zero when the
+// session no longer exists, so this covers "pane gone" and "pane not ready" in one call.
+export function tuiPaneHealthy(tmux, tmuxName) {
+  const r = tmux(["capture-pane", "-p", "-t", tmuxName]);
+  if (!r || r.status !== 0 || typeof r.stdout !== "string") return false;
+  return tuiInputReady(r.stdout);
+}
+
+// Pool pane names carry a "p" marker after the port-scoped prefix:
+//   turn pane: ocp-tui-<port>-<8hex>      (unchanged)
+//   pool pane: ocp-tui-<port>-p<8hex>
+// Purely for operator legibility (`tmux ls` shows which panes are warm). It is NOT the
+// reaper's exemption mechanism — that is the exact-name spare set (see the POOL/REAPER
+// INVARIANT above), so a pooled-LOOKING orphan is still reaped. Both shapes start with
+// sessionPrefixForPort(port), so both remain reapable as "ours", and neither can match
+// LEGACY_SESSION_NAME_RE.
+export function poolPaneName(port, sessionId) {
+  return sessionPrefixForPort(port) + "p" + sessionId.slice(0, 8);
+}
+
+// Boot ONE interactive `claude` pane and wait for its input bar. Shared by the cold
+// request path (runTuiTurn) and the warm pool (lib/tui/pool.mjs) so a pooled pane is
+// spawned with byte-for-byte the same argv, HOME, cwd and trust preparation as a
+// cold-booted one — the pool must not become a second, drifting spawn path.
+//
+// Each pane gets its OWN fresh randomUUID() --session-id, fixed at boot. That is what
+// keeps a pooled pane single-use-safe: its transcript holds exactly one exchange.
+//
+// requireReady: the cold path tolerates a readiness timeout (it falls through and lets
+// the paste-verify decide — pre-existing behaviour, unchanged). The POOL sets it, because
+// a pane that never reached its input bar is worthless as a warm pane and must not be
+// enlisted: throw, let the pool count a bootFailure, and leave the request path to
+// cold-boot as usual.
+// bootMs: max wait for the input bar. Defaults to BOOT_MS (the REQUEST path's cap, which is
+// deliberately tight — a client is blocked on it). The POOL passes POOL_BOOT_MS instead: a
+// background pre-boot has nobody waiting on it, and capping it at the request-path's 4 s
+// made real refills fail (observed live: a refill booting alongside an in-flight turn took
+// >4 s and was discarded, silently lowering the hit rate). Slow != broken for a pre-boot.
+// `sessionId` / `name` (both optional): the caller may supply the pane's identity instead of
+// letting bootTuiPane mint it. The POOL does, because it must know the tmux session's NAME
+// before this function runs — the session is created synchronously below, well before the
+// readiness wait returns, so a pool that only learned the name on resolve could neither spare
+// the session from the reaper nor kill it on shutdown. Supplying BOTH also keeps the name's
+// hex suffix equal to the session-id's, so `tmux ls` correlates to the transcript file.
+// `streamDir` (optional, OCP_TUI_STREAM): install claude's MessageDisplay hook on this pane.
+// Done HERE, at boot — not at turn time — and that is the whole reason streaming survives the
+// WARM POOL: the hook script + settings file are STATIC (one pair per streamDir), and the only
+// per-turn thing, the sink path, is derived from the pane's own --session-id, which is fixed
+// right here. So a pre-booted pane already carries its hook and its own sink and streams exactly
+// like a cold-booted one; nothing request-specific is ever baked into the spawn.
+export async function bootTuiPane({
+  model, claudeBin, home, realHome, cwd, port, entrypointMode = "cli",
+  tmux = defaultTmux, sessionId = null, name = null, requireReady = false, bootMs = BOOT_MS,
+  streamDir = null,
+}) {
+  const sid = sessionId || randomUUID();
+  // Port-scoped session name (F7 fix) — see sessionPrefixForPort / reapStaleTuiSessions
+  // for why this instance's own listen port is the namespace discriminator.
+  const tmuxName  = name || (sessionPrefixForPort(port) + sid.slice(0, 8));
+  const ehome     = home || process.env.HOME;       // HOME claude runs under (scratch or real)
+  const rhome     = realHome || process.env.HOME;   // real home (OAuth + onboarded config source)
+
+  // Env-token-only mode: the env token is set AND claude runs in an isolated home
+  // (ehome !== rhome). In that case the scratch home must be CREDENTIAL-FREE (no
+  // .credentials.json) so the env token — passed by buildTuiCmd — is the only credential
+  // and is therefore authoritative (interactive claude otherwise PREFERS a credentials.json,
+  // shadowing the env token; proven live on PI231). server.mjs derives TUI_HOME via
+  // resolveTuiHome() so this isolated home is the DEFAULT once CLAUDE_CODE_OAUTH_TOKEN is set.
+  const envTokenMode = !!process.env.CLAUDE_CODE_OAUTH_TOKEN && ehome !== rhome;
+
+  // Ensure scratch cwd exists, then prepare the (scratch or real) HOME + trust the
+  // cwd — before claude boots.
+  if (!existsSync(cwd)) mkdirSync(cwd, { recursive: true });
+  prepareTuiHome(rhome, ehome, cwd, { envTokenMode });
+
+  // Streaming sink for THIS pane (see the streamDir note above). rmSync first so a
+  // re-used session-id can never replay a previous turn's deltas.
+  let streamFile = null, streamSettings = null;
+  if (streamDir) {
+    streamFile     = streamFilePath(streamDir, sid);
+    streamSettings = prepareStreamHook(streamDir);
+    try { rmSync(streamFile, { force: true }); } catch { /* start from a fresh sink */ }
+  }
+
+  // Minimal env for spawnSync (tmux itself). The pane's claude env comes exclusively
+  // from the `env` prefix string built inside buildTuiCmd — tmux does NOT forward the
+  // spawning process's env to the pane, so the {env} here is intentionally minimal.
+  const env = { ...process.env };
+  env.HOME = ehome; // tmux needs HOME; all claude-specific vars go via buildTuiCmd prefix
+
+  // Boot the interactive session inside tmux, rooted at the scratch cwd.
+  // Capture the result: if tmux new-session fails (status !== 0) there is no PTY, no
+  // interactive spawn — abort BEFORE the boot wait rather than paste into a non-existent
+  // session or issue a billing request without a verified interactive context.
+  const spawnResult = tmux(
+    ["new-session", "-d", "-s", tmuxName, "-x", "220", "-y", "50", "-c", cwd,
+     buildTuiCmd(claudeBin, model, sid, ehome, entrypointMode,
+                 streamFile ? { file: streamFile, settings: streamSettings } : null)],
+    { env },
+  );
+  if (!spawnResult || spawnResult.status !== 0) {
+    throw new Error("tui_spawn_failed: tmux session not created");
+  }
+
+  // Wait until claude's input bar is actually ready (not a blind sleep).
+  // bootMs is the MAX readiness wait, not a fixed delay.
+  const ready = await pollUntil(() => tuiInputReady(tuiCapturePane(tmux, tmuxName)),
+    { timeoutMs: bootMs, intervalMs: READY_POLL_MS });
+  if (!ready) {
+    if (requireReady) {
+      try { tmux(["kill-session", "-t", tmuxName]); } catch { /* already gone */ }
+      throw new Error("tui_pane_not_ready: input bar did not appear within " + bootMs + "ms");
+    }
+    // Cold path (pre-existing behaviour): readiness timed out; rely on paste-verify.
+    console.error("[tui] input_not_ready", tmuxName);
+  }
+  return { name: tmuxName, sessionId: sid, model, ehome, streamFile, bootedAt: Date.now() };
+}
+
+// Full per-request TUI lifecycle:
+//   1. Take a WARM pane from the pool if one is available for this model (opt-in;
+//      OCP_TUI_POOL_SIZE=0 => always null => steps 2-3 below are exactly today's path).
+//      A pooled pane is SINGLE-USE: it already carries its own fresh --session-id, it
+//      serves this one turn, and it is killed in the finally like any other pane.
+//   2. On a MISS: pre-trust the scratch cwd, boot an interactive `claude` in a fresh tmux
+//      session in the scratch cwd, poll capture-pane until the `? for shortcuts` input bar
+//      appears (bootTuiPane). BOOT_MS is the max wait, not a fixed delay.
+//   3. Write prompt to a 0600 temp file (no shell injection from prompt content).
+//   4. Paste the prompt via tmux load-buffer + paste-buffer -p (bracketed paste) —
+//      reliable for large multi-line prompts where send-keys -l is not (issue #130).
+//      Poll-verify the prompt landed in the input (placeholder gone / [Pasted text]);
+//      fast-fail with tui_paste_not_landed if it never lands (prevents the 120s
+//      wallclock "stuck typing" hang). Then submit with a SEPARATE Enter key event.
+//   5. Block on the native JSONL transcript (located by THIS pane's session-id) until
+//      terminal marker or wall-clock cap.
+//   6. Always teardown: kill session + rm temp dir (even on throw), and kick a background
+//      pool refill so the next request finds a warm pane.
+// Returns { text, entrypoint } from readTuiTranscript (entrypoint is the billing-pool
+// classifier, e.g. "cli", or null if the transcript did not include a turn_duration).
+//
+// STREAMING (OCP_TUI_STREAM, default off). Pass `onDelta` and `streamDir`, and the pane's
+// MessageDisplay hook (installed by bootTuiPane; see lib/tui/stream.mjs) appends each raw
+// delta payload to the pane's own sink. This driver polls that sink and invokes onDelta(payload)
+// per fire while the turn is still generating. A WARM pane already carries its sink from boot
+// (pane.streamFile), so the pooled and cold paths stream identically.
+//
+// `streamDir` IS PASSED TO THE COLD BOOT UNCONDITIONALLY (not gated on `onDelta`) — F4 fix. The
+// spawn argv is this project's billing-classification surface: a caller with OCP_TUI_STREAM on
+// but THIS particular request non-streaming (stream:false) must still get the SAME argv whether
+// it lands on a pool HIT or a cold-boot MISS, because a pre-booted pool pane cannot know in
+// advance whether the request it will eventually serve wants streaming — it installs the hook
+// unconditionally whenever the pool is warming at all (see server.mjs's bootPane closure). Gating
+// the cold boot's hook install on `onDelta` made a stream:false request's argv depend on whether
+// it happened to hit the pool or miss it — the exact drift this surface cannot tolerate. Whether
+// the hook is actually POLLED is a separate, correctly-scoped decision: see `streaming` below,
+// gated on onDelta && streamFile, so a non-streaming turn never reads its own sink even though
+// the hook is running.
+//
+// The transcript stays AUTHORITATIVE regardless: it is still the terminal-turn signal, still the
+// source of the returned `text`, and still the input to the caller's honesty gates. The delta
+// stream is a low-latency MIRROR of it, never a replacement, and the caller asserts the two
+// agree. With onDelta AND streamDir both omitted, nothing here changes: no poll, no hook.
+//
+// `abortSignal` (optional): aborts the transcript wait, so a client that disconnects mid-turn
+// tears the pane down NOW (the finally below) instead of holding the pane — and therefore the
+// caller's semaphore slot — until the turn or the wallclock cap ends.
+export async function runTuiTurn({
+  prompt,
+  model,
+  claudeBin,
+  home,
+  realHome,
+  cwd,
+  port,
+  wallclockMs = 120000,
+  entrypointMode = "cli",
+  tmux = defaultTmux,
+  pool = null,       // TuiPanePool | null — null (default) === today's cold-boot-only path
+  onPane = null,     // optional observer: ({ warm }) => void, for logging/metrics
+  onDelta = null,    // (payload) => void — invoked per MessageDisplay hook fire, mid-turn
+  streamDir = null,  // hook sink dir, passed to the COLD boot UNCONDITIONALLY (F4 — see above);
+                      // a warm pane brings its own, fixed at its own boot
+  abortSignal = null,
+}) {
+  // 1. Warm pane, or cold boot. A MISS is never an error — it is exactly today's path.
+  let pane = pool ? pool.acquire(model) : null;
+  const warm = !!pane;
+  // Kick the refill IMMEDIATELY (not after the turn): the replacement pane then boots
+  // CONCURRENTLY with this turn and is warm by the time the next request arrives. Also
+  // runs on a MISS — acquire() has just retargeted the pool to this model, so the miss
+  // that cold-boots today warms the pool for the next caller. Fire-and-forget; it takes
+  // no TuiSemaphore slot (see pool.refill's SLOT ACCOUNTING note).
+  if (pool) pool.refill();
+  if (onPane) { try { onPane({ warm }); } catch { /* observer must never break a turn */ } }
+  if (!pane) {
+    // streamDir passed AS-IS (not gated on onDelta) — F4: see the STREAMING comment above.
+    pane = await bootTuiPane({ model, claudeBin, home, realHome, cwd, port, entrypointMode, tmux,
+                              streamDir });
+  }
+  const tmuxName  = pane.name;
+  const sessionId = pane.sessionId;   // THIS pane's own session-id — one session, one turn
+  const ehome     = pane.ehome || home || process.env.HOME;
+
+  // Streaming state is read off the PANE, not recomputed here — a warm pane fixed its sink at
+  // boot, and a cold one just did the same above. If the pool was booted WITHOUT a streamDir
+  // while onDelta is set, streamFile is null and the turn degrades to buffered: correct, just
+  // not fast. (server.mjs wires the same streamDir into both paths so that cannot happen.)
+  const streamFile = pane.streamFile || null;
+  const streaming  = !!(onDelta && streamFile);
+  const streamCursor = { consumed: 0 };
+  let streamStopped = false;
+  let pollTimer = null;
+  // Drain every complete line appended since the last drain. Never throws into the turn: a
+  // malformed line is skipped by parseDeltaChunk, and an onDelta that throws is contained.
+  const drainDeltas = () => {
+    if (!streaming) return;
+    let text;
+    try { text = readFileSync(streamFile, "utf8"); } catch { return; } // absent until the first fire
+    const { deltas, consumed } = parseDeltaChunk(text, streamCursor.consumed);
+    streamCursor.consumed = consumed;
+    for (const d of deltas) {
+      try { onDelta(d); } catch { /* a sink error must never abort the turn */ }
+    }
+  };
+
+  // Write prompt to a temp file (mode 0600) so the content never touches argv.
+  const tmpDir     = mkdtempSync(`${tmpdir()}/ocp-tui-`);
+  const promptFile = `${tmpDir}/prompt.txt`;
+  writeFileSync(promptFile, prompt, { mode: 0o600 });
+
+  try {
+    // 3. Paste the prompt via a tmux PASTE BUFFER with bracketed paste (-p), NOT
+    //    `send-keys -l`. send-keys of a large multi-line prompt is unreliable: the
+    //    embedded newlines arrive as separate key events (effectively repeated Enter),
+    //    so a big OpenClaw-style prompt never lands and the turn hangs to the wallclock
+    //    (issue #130 — reproduced at ~300 lines; fixed by bracketed paste). load-buffer
+    //    reads the file directly (no shell arg limit, no `"$(cat)"`), and paste-buffer -p
+    //    wraps it in bracketed-paste markers so claude ingests it atomically as ONE paste
+    //    ("[Pasted text #N +M lines]"). -d deletes the buffer afterward. Buffer name is the
+    //    per-session tmuxName, so concurrent turns never collide.
+    tmux(["load-buffer", "-b", tmuxName, promptFile]);
+    tmux(["paste-buffer", "-b", tmuxName, "-t", tmuxName, "-p", "-d"]);
+
+    // Verify the prompt POSITIVELY landed before submitting; poll (a large bracketed paste
+    // takes a beat to render the "[Pasted text]" indicator). This is load-bearing: firing
+    // Enter before the paste renders submits an empty box → the turn hangs to the wallclock
+    // (issue #130). Fast-fail if it never lands → deterministic error in seconds.
+    const landed = await pollUntil(() => tuiPromptLanded(tuiCapturePane(tmux, tmuxName), prompt),
+      { timeoutMs: PASTE_VERIFY_MS, intervalMs: READY_POLL_MS });
+    if (!landed) {
+      throw new Error("tui_paste_not_landed: prompt did not reach claude's input within " + PASTE_VERIFY_MS + "ms");
+    }
+
+    // Submit (separate Enter key event).
+    tmux(["send-keys", "-t", tmuxName, "Enter"]);
+
+    // 5a. Streaming only: start polling the hook sink. Runs CONCURRENTLY with the
+    //     transcript wait below — the deltas are what make the answer visible while the
+    //     turn is still generating; the transcript is what makes it authoritative.
+    if (streaming) {
+      const loop = () => {
+        if (streamStopped) return;
+        drainDeltas();
+        pollTimer = setTimeout(loop, STREAM_POLL_MS);
+      };
+      pollTimer = setTimeout(loop, STREAM_POLL_MS);
+    }
+
+    // 5b. Block on the native transcript (resolved by THIS pane's session-id) until terminal.
+    //     Returns { text, entrypoint, truncated } from readTuiTranscript.
+    const result = await readTuiTranscript({ home: ehome, sessionId, wallclockMs, abortSignal });
+
+    // 5c. FINAL drain. The terminal marker can land between two poll ticks, so the last
+    //     delta(s) may still be unread — without this the tail would be missing from the
+    //     stream and every turn would need a transcript top-up.
+    streamStopped = true;
+    if (pollTimer) clearTimeout(pollTimer);
+    drainDeltas();
+    return result;
+  } finally {
+    // 6. Teardown — always, even on throw (including an abortSignal disconnect, which is
+    //    exactly why the pane cannot outlive a client that walked away). A pooled pane is
+    //    torn down here exactly like a cold-booted one: SINGLE-USE, never returned (pool.mjs).
+    streamStopped = true;
+    if (pollTimer) clearTimeout(pollTimer);
+    try { tmux(["kill-session", "-t", tmuxName]); } catch { /* already gone */ }
+    try { rmSync(tmpDir, { recursive: true, force: true }); } catch { /* best effort */ }
+    if (streamFile) { try { rmSync(streamFile, { force: true }); } catch { /* best effort */ } }
+  }
+}

lib/tui/stream.mjs

@@ -0,0 +1,288 @@
+// TUI-mode real SSE streaming — the `MessageDisplay` hook sink.
+//
+// WHAT THIS IS. `claude` fires a **MessageDisplay** hook per rendered block of the
+// assistant's reply, handing the hook the RAW MARKDOWN SOURCE of an incremental
+// `delta` on stdin. Registered via `--settings` on the ordinary interactive TUI spawn
+// (NO -p, NO --bare — the billing pool is untouched), it is the only byte-faithful
+// incremental source the interactive CLI exposes. Everything here consumes that hook
+// surface AS EMITTED — forwarding, not inventing.
+//
+// ALIGNMENT.md: **Class B**. We consume claude's own hook payload and re-emit it in the
+// OpenAI chat/completions streaming shapes OCP already speaks (ADR 0006). There is no
+// `cli.js` citation because no `cli.js` function is being mirrored: the TUI spawn is
+// OCP-owned surface (ADR 0007), and the hook payload is claude's own published contract.
+//
+// THE VERIFIED CONTRACT (docs/plans/2026-07-13-tui-latency/streaming-spike.md, and
+// independently reproduced on claude 2.1.207 / sonnet-4-6 / banner `· Claude Max`):
+//
+//   payload (stdin, one JSON object per fire):
+//     { hook_event_name:"MessageDisplay", session_id, transcript_path, prompt_id, cwd,
+//       turn_id, message_id, index, final, delta }
+//
+//   - deltas carry the raw markdown source (`## `, `**`, ```javascript all present)
+//   - concat(deltas of one message) === T, byte-exactly   (T = extractLatestAssistantText)
+//   - T.startsWith(concat(deltas[0..n])) at EVERY n       (prefix-stable)
+//   - block-level granularity (~5-7 fires per answer), NOT token-level
+//   - only `text` blocks fire it — thinking blocks are excluded (what OCP wants)
+//
+// ⚠️ THE HOOK IS SYNCHRONOUS. The hook's source sets `forceSyncExecution: true` —
+// `claude` BLOCKS on every fire. The hook script must therefore write and exit, doing
+// NO work inline. Measured cost of the script below: p50 7.2 ms / p90 14.7 ms per fire,
+// i.e. ~50 ms added blocking across a whole ~7-delta turn against a 6-10 s turn. That is
+// noise, so a plain append is the right sink — a FIFO would be faster on paper but a FIFO
+// blocks its writer until a reader attaches, which would hand `claude` a way to hang.
+//
+// WARM-POOL COMPATIBILITY (load-bearing — a warm pane pool is a separate in-flight PR).
+// The hook script and the settings file are BOTH STATIC: one copy per stream dir, written
+// once, never per-request. The per-turn destination is carried in the PANE'S OWN ENV as
+// `OCP_TUI_STREAM_FILE` (verified live: a hook inherits the pane's environment), and the
+// path is derived from the session-id — which for a pre-booted pane is fixed at BOOT.
+// Nothing about a request is baked into the settings file at spawn time, so a pane booted
+// before its request arrives streams exactly the same way.
+import { writeFileSync, mkdirSync, renameSync } from "node:fs";
+import { detectTuiUpstreamError } from "./transcript.mjs";
+
+// Default holdback before the first byte is released to the client. See TuiDeltaAssembler.
+export const DEFAULT_HOLDBACK_CHARS = 100;
+
+// Resolve OCP_TUI_STREAM_HOLDBACK to a SAFE value. The whole C-1 auth-banner guarantee rests
+// on the holdback being at least the default banner detector's max message length — which is
+// exactly DEFAULT_HOLDBACK_CHARS. So this is a FLOOR, not a hint: a smaller value (or a NaN
+// typo like "unlimited"/"5MB") would let a real banner fragment release before the terminal
+// detector could classify the whole message, silently reopening the leak the assembler exists
+// to prevent. The env var's own doc says "Only raise it"; this enforces that instead of trusting
+// it. Returns { value, clamped } so the caller can warn when it had to clamp — a silent floor is
+// less honest than a noticed one.
+export function resolveStreamHoldback(raw, floor = DEFAULT_HOLDBACK_CHARS) {
+  const parsed = parseInt(raw ?? "", 10);
+  if (!Number.isFinite(parsed)) return { value: floor, clamped: raw != null && String(raw).trim() !== "" };
+  if (parsed < floor) return { value: floor, clamped: true };
+  return { value: parsed, clamped: false };
+}
+
+// The hook script. POSIX sh, no interpreter startup beyond /bin/sh, one fork (`cat`).
+//
+//   - `printf` is a shell BUILTIN in sh/dash/bash, so the newline costs no fork.
+//   - the `{ cat; printf '\n'; } >>` group opens the file ONCE and appends both writes
+//     through the same O_APPEND fd, so a payload and its terminator can never be split
+//     by another writer. (They never race anyway: one file per pane, and MessageDisplay
+//     is synchronous within a pane.)
+//   - a payload JSON can never contain a literal newline — JSON.stringify escapes them —
+//     so "one line == one payload" holds, and a torn write is always a trailing partial
+//     line, which parseDeltaChunk() leaves unconsumed until it completes.
+//   - NO OCP_TUI_STREAM_FILE (e.g. a pane booted with streaming off, or any other claude
+//     session that happens to load this settings file) => swallow stdin and exit 0. The
+//     hook must NEVER fail or block: claude is waiting on it.
+export const HOOK_SCRIPT = `#!/bin/sh
+# OCP TUI streaming sink — claude fires this per MessageDisplay block and BLOCKS on it.
+# Write and exit. Never do work here.
+[ -n "\$OCP_TUI_STREAM_FILE" ] || exec cat >/dev/null
+{ cat; printf '\\n'; } >> "\$OCP_TUI_STREAM_FILE"
+`;
+
+// The --settings payload registering the hook. Static: no per-request data.
+export function buildStreamSettings(hookScriptPath) {
+  return { hooks: { MessageDisplay: [{ hooks: [{ type: "command", command: hookScriptPath }] }] } };
+}
+
+export const hookScriptPath   = (streamDir) => `${streamDir}/md-hook.sh`;
+export const streamSettingsPath = (streamDir) => `${streamDir}/settings.json`;
+// One file per session-id. For a pre-booted (warm) pane the session-id is fixed at boot,
+// so this path is knowable at boot — which is what keeps the pool compatible.
+export const streamFilePath   = (streamDir, sessionId) => `${streamDir}/${sessionId}.jsonl`;
+
+// Atomic write: temp file + rename (same-directory, same-filesystem, so rename is atomic on
+// POSIX). A process killed mid-`writeFileSync` leaves the TEMP file half-written, never the
+// real path — `path` always names either the old complete content or the new complete
+// content, never a torn one. That matters specifically for md-hook.sh: it is SYNCHRONOUS
+// (claude blocks on every fire), so a truncated script would still pass `existsSync`, still
+// get exec'd, and fail/hang on every single MessageDisplay fire with no operator-visible
+// symptom short of streaming going silently dead (F7's streamZeroDeltaTurns is the backstop
+// for exactly that). Mirrors ensureTuiCwdTrusted's tmp+renameSync pattern in session.mjs.
+function writeFileAtomic(path, content, mode) {
+  const tmp = `${path}.${process.pid}.tmp`;
+  writeFileSync(tmp, content, { mode });
+  renameSync(tmp, path);
+}
+
+// Write the static hook script + settings file into `streamDir`. UNCONDITIONAL, not
+// write-if-missing: these files persist across OCP restarts at `streamDir`, so a host that
+// booted once under an older version and never had its stream dir cleared would otherwise be
+// silently stuck on a stale HOOK_SCRIPT / buildStreamSettings() forever — no future OCP
+// upgrade could ever reach it. Safe to call every boot: the content is static (no per-request
+// data), so a same-content rewrite is the overwhelmingly common case and costs two tiny
+// atomic writes, not a per-turn expense. Returns the settings path to hand to `claude
+// --settings`.
+export function prepareStreamHook(streamDir) {
+  mkdirSync(streamDir, { recursive: true });
+  const script = hookScriptPath(streamDir);
+  const settings = streamSettingsPath(streamDir);
+  writeFileAtomic(script, HOOK_SCRIPT, 0o700);
+  writeFileAtomic(settings, JSON.stringify(buildStreamSettings(script), null, 2), 0o600);
+  return settings;
+}
+
+// Parse newly-appended sink lines. `consumed` is the number of COMPLETE lines already
+// taken; only lines terminated by "\n" are complete, so a payload caught mid-write stays
+// unconsumed until its terminator lands. Returns the fresh MessageDisplay payloads plus
+// the new consumed count. Pure — the caller owns the cursor.
+export function parseDeltaChunk(text, consumed = 0) {
+  const lines = String(text ?? "").split("\n");
+  const complete = lines.slice(0, -1); // the tail after the last "\n" is a partial line
+  const deltas = [];
+  for (const line of complete.slice(consumed)) {
+    const t = line.trim();
+    if (!t) continue;
+    try {
+      const o = JSON.parse(t);
+      if (o && o.hook_event_name === "MessageDisplay" && typeof o.delta === "string") deltas.push(o);
+    } catch { /* not ours / not parseable — skip, never throw into the request path */ }
+  }
+  return { deltas, consumed: complete.length };
+}
+
+// ── The assembler: hook deltas → client bytes, with the honesty gates intact ──
+//
+// Two jobs, both load-bearing.
+//
+// 1. THE AUTH-BANNER HOLDBACK (C-1 / issue #133 must survive streaming).
+//    The interactive CLI renders an auth failure as ordinary assistant TEXT — so an
+//    expired-credential turn fires MessageDisplay with the BANNER as its delta, and a
+//    naive forwarder would stream "Please run /login · API Error: 401 …" to the client as
+//    a normal answer, exactly the silent-error case C-1 exists to prevent.
+//    detectTuiUpstreamError() classifies a WHOLE message, so it cannot be run per-delta.
+//    Instead we HOLD BACK the first `holdbackChars` characters. The default detector only
+//    ever fires on a message of <= 100 chars (TUI_ERR_MAX_LEN — real banners are 69 and 73),
+//    so once the TRIMMED accumulation EXCEEDS 100 chars the final text cannot be a banner by
+//    that detector's own length rule, and releasing is safe. An answer that never exceeds the
+//    holdback is simply delivered whole at terminal — i.e. exactly today's buffered
+//    behaviour, gates and all.
+//    THE GUARANTEE HAS TWO HALVES, both required — neither alone is sufficient:
+//      (i)  Nothing is emitted for a message until its trimmed accumulation exceeds the
+//           detector's max banner length. This is what keeps the FIRST message of a turn
+//           safe: a banner-length message can never clear the holdback.
+//      (ii) Once a message boundary follows an emit (`restartedAfterEmit`), push() stops
+//           emitting ENTIRELY for the rest of the turn — a SECOND message (e.g. an
+//           auth-failure banner rendered mid-turn, after tool-using prose already streamed)
+//           gets zero bytes forwarded, not just a fresh holdback of its own. finalize() then
+//           refuses the whole turn (SSE error frame, no cache) precisely because the first
+//           message's bytes are unretractable and unverifiable against T. Without this half,
+//           (i) alone only protects the FIRST message per turn — see F1.
+//    ⚠️ Soundness is w.r.t. the DEFAULT detector. An operator who REPLACES it via
+//    CLAUDE_TUI_ERROR_PATTERNS with a pattern that can match a longer message must raise
+//    OCP_TUI_STREAM_HOLDBACK past their longest banner; server.mjs warns at boot. That is the
+//    one case (i) does not cover — (ii) still applies regardless. Even past both, the
+//    terminal gate still refuses to cache a banner and still ends the stream on an SSE error
+//    frame rather than finish_reason:"stop" — the holdback is the first of two layers, not
+//    the only one.
+//
+// 2. MESSAGE SCOPING (keeps `concat === T` the RIGHT assertion).
+//    The transcript's T is extractLatestAssistantText() — the LAST text-bearing assistant
+//    entry, not every assistant entry. A tool-using turn therefore has TWO messages
+//    (prose → tool_use → answer) and T is only the second. So the assembler scopes to the
+//    CURRENT message_id: when a new message_id appears and NOTHING has been emitted yet,
+//    the held text is DISCARDED — the transcript is about to discard it too, so this keeps
+//    us byte-identical to the buffered path instead of streaming prose the buffered path
+//    would have dropped. When a new message_id appears AFTER we have already emitted, the
+//    bytes are gone and cannot be retracted: finalize() then reports !ok and the caller
+//    fails the turn loudly (SSE error frame, no cache, counted on /health). Fail-loud is
+//    the correct posture — a proxy that silently serves text the transcript disagrees with
+//    is the exact class of bug ALIGNMENT.md exists to prevent.
+// Sentinel for "no message seen yet". Deliberately not null/undefined — see the constructor.
+const NO_MESSAGE_YET = Symbol("no-message-yet");
+
+export class TuiDeltaAssembler {
+  constructor({ holdbackChars = DEFAULT_HOLDBACK_CHARS, detectError = detectTuiUpstreamError } = {}) {
+    this.holdbackChars = holdbackChars;
+    this.detectError = detectError;
+    this.emitted = "";   // bytes ALREADY written to the client — unretractable
+    this.pending = "";   // held back, not yet written
+    this.released = false;
+    // NOT null: a payload may legitimately carry message_id === null, and if the sentinel were
+    // also null the FIRST such payload would compare equal to it, register no boundary, and
+    // leave `messages` at 0 — which used to disarm the restartedAfterEmit guard below entirely.
+    // A unique object is === to nothing a JSON payload can produce, so the first fire ALWAYS
+    // registers as message 1, whatever its message_id is (or isn't).
+    this.messageId = NO_MESSAGE_YET;
+    this.deltas = 0;     // hook fires seen
+    this.messages = 0;   // distinct message_ids seen
+    this.restartedAfterEmit = false;
+  }
+
+  // All hook bytes for the CURRENT message (emitted + still held).
+  get full() { return this.emitted + this.pending; }
+
+  // Feed one MessageDisplay payload. Returns the text to emit NOW, or null (held back).
+  push(payload) {
+    const delta = payload && typeof payload.delta === "string" ? payload.delta : "";
+    const mid = payload ? payload.message_id : null;
+    if (mid !== this.messageId) {
+      this.messageId = mid;
+      this.messages++;
+      if (this.emitted === "") {
+        this.pending = "";        // safe: the transcript will drop this message too
+      } else {
+        // A boundary while bytes are ALREADY out is unrecoverable, full stop — the count of
+        // messages seen so far is irrelevant. The old `else if (this.messages > 1)` guard was
+        // the sole reason a null-message_id first payload could disarm F1: it left `messages`
+        // at 0, so the real boundary evaluated 1 > 1 === false and never armed. The invariant
+        // is "a boundary occurred while emitted !== ''", and that is exactly what this says.
+        this.restartedAfterEmit = true; // unrecoverable — finalize() will refuse the turn
+      }
+    }
+    this.deltas++;
+    // F1: once a message boundary has followed an emit, the turn is ALREADY unrecoverable —
+    // finalize() will refuse it (see restartedAfterEmit above). `this.released` stays true
+    // from the FIRST message's release and, uncorrected, lets every later message's deltas
+    // stream straight through unfiltered — exactly the auth-banner-mid-turn leak this class
+    // exists to prevent. Stop emitting HERE, permanently, for the rest of the turn: there is
+    // nothing left to gain from continuing to forward bytes for a turn that will be refused,
+    // and every byte forwarded now is one more the client cannot be told to un-see.
+    if (this.restartedAfterEmit) return null;
+    if (!delta) return null;
+
+    if (this.released) {
+      this.emitted += delta;
+      return delta;
+    }
+    this.pending += delta;
+    // Release only once the TRIMMED accumulation is past the banner detector's reach.
+    // detectTuiUpstreamError() trims before measuring length (TUI_ERR_MAX_LEN is a trimmed-
+    // length bound), so gating release on the UNTRIMMED pending.length let a run of >
+    // holdbackChars whitespace trim down to "" — detectError("") sees nothing to classify,
+    // returns null, and release fires with the holdback never having actually screened
+    // anything. Trimming here keeps both sides of the check talking about the same string.
+    if (this.pending.trim().length > this.holdbackChars && this.detectError(this.pending) == null) {
+      const out = this.pending;
+      this.pending = "";
+      this.released = true;
+      this.emitted += out;
+      return out;
+    }
+    return null;
+  }
+
+  // Reconcile against the AUTHORITATIVE transcript text T. Call only AFTER the truncation
+  // and auth-banner gates have passed. Returns:
+  //   { ok:true,  tail, exact }  — tail is the remaining text to emit (may be ""). `exact`
+  //                               is concat(deltas) === T; when false we still serve exactly
+  //                               T, having topped up from the transcript, and the caller
+  //                               counts a topUp.
+  //   { ok:false, ... }          — what we already emitted is NOT a prefix of T. The client
+  //                               holds bytes the transcript disagrees with; the caller must
+  //                               NOT cache and must end the stream on an SSE error frame.
+  finalize(T) {
+    const text = typeof T === "string" ? T : "";
+    const full = this.full;
+    if (!text.startsWith(this.emitted)) {
+      return { ok: false, tail: null, exact: false, emitted: this.emitted.length, transcript: text.length };
+    }
+    return {
+      ok: true,
+      tail: text.slice(this.emitted.length),
+      exact: full === text,
+      emitted: this.emitted.length,
+      transcript: text.length,
+    };
+  }
+}

lib/tui/transcript.mjs

@@ -0,0 +1,301 @@
+// Transcript reader for TUI-mode. Reads claude's native JSONL session transcript
+// and returns the latest assistant turn's text once the turn is terminal.
+//
+// Authority: claude CLI v2.1.157 — interactive session transcript at
+//   <HOME>/.claude/projects/<CWD with every "/" -> "-">/<--session-id>.jsonl
+// Completion marker: a line {"type":"system","subtype":"turn_duration",...}.
+// See docs/superpowers/specs/2026-05-30-tui-mode-production-design.md §4.
+import { readFileSync, existsSync, readdirSync } from "node:fs";
+
+const sleep = (ms) => new Promise((r) => setTimeout(r, ms));
+
+// Locate a session's transcript by its UUID across every projects subdir, without
+// reconstructing the encoded cwd. Robust to whatever encoding claude applies.
+// Returns the path, or null if not present yet (it appears once the turn starts).
+// TODO: add a CI fixture-contract test (a captured real transcript) so schema drift
+//       in the claude JSONL format fails loudly rather than silently degrading.
+export function findTranscriptPath(home, sessionId) {
+  if (!home || !sessionId) return null;
+  const root = `${home}/.claude/projects`;
+  let dirs;
+  try { dirs = readdirSync(root); } catch { return null; }
+  for (const d of dirs) {
+    const candidate = `${root}/${d}/${sessionId}.jsonl`;
+    if (existsSync(candidate)) return candidate;
+  }
+  return null;
+}
+
+// Parse NDJSON text into objects; skip blank lines and partial/forming lines
+// (the live transcript is read mid-write, so the last line may be incomplete).
+export function parseTranscriptLines(text) {
+  const out = [];
+  for (const line of text.split("\n")) {
+    const t = line.trim();
+    if (!t) continue;
+    try { out.push(JSON.parse(t)); } catch { /* partial line being written */ }
+  }
+  return out;
+}
+
+// A line marks the assistant turn complete when EITHER:
+//   (a) {type:"system", subtype:"turn_duration"} — emitted by newer claude builds
+//       (e.g. 2.1.159), OR
+//   (b) {type:"assistant"} whose message.stop_reason is a FINAL reason
+//       ("end_turn" / "stop_sequence" / "max_tokens"). This is the API-level
+//       end-of-turn signal, present across claude builds whose transcripts do NOT
+//       emit turn_duration (e.g. 2.1.114 — verified live on the cloud host). Without
+//       it OCP can't detect completion on those builds and hangs to the wallclock,
+//       then returns only partial text (issue #130, cloud/server-side symptom).
+//
+// stop_reason "tool_use" is deliberately NOT terminal: the model is mid-turn (it will
+// run a tool and continue with a later assistant entry). Matching on a FINAL
+// stop_reason — not on the mere presence of a tool_use — keeps tool-using turns intact.
+// (The v3.17.1 narrowing dropped a buggy "tool_use is terminal" rule; this restores
+// cross-version completion detection without bringing that bug back.)
+const TERMINAL_STOP_REASONS = new Set(["end_turn", "stop_sequence", "max_tokens"]);
+export function isTerminalLine(obj) {
+  if (!obj || typeof obj !== "object") return false;
+  if (obj.type === "system" && obj.subtype === "turn_duration") return true;
+  if (obj.type === "assistant" && obj.message && typeof obj.message === "object") {
+    return TERMINAL_STOP_REASONS.has(obj.message.stop_reason);
+  }
+  return false;
+}
+
+// Text of the LAST assistant turn: concatenate its text content blocks
+// (ignore thinking/tool_use blocks). Later assistant entries overwrite earlier.
+// Fixture-confirmed shape: top-level type:"assistant", message.content[] array.
+//
+// Scoping: this returns the FINAL text-bearing assistant entry in the whole file,
+// not "text since the matching user line" (spec §4.2). Those are equivalent ONLY
+// under OCP's one-session-per-request model (a fresh --session-id => a fresh
+// transcript holding one logical exchange). If a future warm-pool ever reuses a
+// session WITHOUT a fresh session-id / clear, earlier-turn text could leak — that
+// author must add user-line scoping here. See spec §7.2.
+//
+// STATUS (warm pool, lib/tui/pool.mjs — the "future warm-pool" this note anticipated):
+// the pool does NOT reuse sessions, so the precondition above still holds and no
+// user-line scoping was added. Each pooled pane is booted with its OWN fresh
+// randomUUID() --session-id (bootTuiPane) and is SINGLE-USE: it serves exactly one turn
+// and is then killed and replaced. One session still means one logical exchange, so the
+// last assistant entry is still that request's answer.
+// The warning therefore stands UNCHANGED for anyone who later wants a pane to serve a
+// SECOND turn (or to reset one with /clear and reuse it): that is a leak, and it needs
+// user-line scoping HERE before it can be safe. Do not relax pool.mjs's single-use rule
+// without doing that work first.
+export function extractLatestAssistantText(events) {
+  let text = "";
+  for (const ev of events) {
+    if (!ev || ev.type !== "assistant") continue;
+    const content = ev.message && ev.message.content;
+    if (!Array.isArray(content)) continue;
+    const parts = content
+      .filter((b) => b && b.type === "text" && typeof b.text === "string")
+      .map((b) => b.text);
+    if (parts.length) text = parts.join("");
+  }
+  return text;
+}
+
+// Returns the entrypoint string (e.g. "cli") used for the billing-pool assertion,
+// or null if absent. Lets callers assert the subscription-classified path.
+//
+// Resolution order (C-3, issue #133):
+//   1. PREFER the turn_duration system line's `entrypoint` — the authoritative
+//      end-of-turn classifier emitted by builds that produce turn_duration
+//      (e.g. claude-2.1.104/2.1.157 on PI231).
+//   2. FALL BACK to the `entrypoint` field on ANY ordinary transcript line
+//      (assistant / user / attachment / system) — present on BOTH emitting and
+//      non-emitting builds. Some claude builds (e.g. certain Mac mini transcripts)
+//      do NOT emit a turn_duration line at all; reading ONLY turn_duration made the
+//      caller's tui_entrypoint_mismatch assertion (server.mjs) get got:null every
+//      turn and go blind. The entrypoint value is identical across line types within
+//      a single interactive session (fixture-confirmed: every line in
+//      complete-haiku.jsonl carrying `entrypoint` reads "cli"), so the fallback
+//      yields the same classifier. Last-writer-wins on the fallback.
+export function verifyEntrypoint(events) {
+  let fallback = null;
+  for (const ev of events) {
+    if (!ev || typeof ev !== "object") continue;
+    if (ev.type === "system" && ev.subtype === "turn_duration" && ev.entrypoint != null) {
+      return ev.entrypoint; // authoritative — short-circuit
+    }
+    if (ev.entrypoint != null) fallback = ev.entrypoint;
+  }
+  return fallback;
+}
+
+// ── C-1: honest AUTH-FAILURE banner detection (issue #133) ───────────────
+// When the interactive `claude` CLI hits an in-session error it does NOT crash —
+// it renders the error as ordinary assistant text in the transcript. The specific
+// failure C-1 exists to catch is R-1: EXPIRED / INVALID credentials, where every
+// turn comes back as the same one-line auth-failure banner and OCP, none the wiser,
+// caches that banner (server.mjs setCachedResponse), shares it via singleflight, and
+// records a model SUCCESS — so a hard auth error is silently served (and cached for
+// the 5-min TTL) as a real answer. The two live-reproduced banners on PI231
+// (2026-06-10) are:
+//   "Please run /login · API Error: 401 Invalid authentication credentials"  (69 chars)
+//   "Failed to authenticate. API Error: 401 Invalid authentication credentials" (73 chars)
+//
+// WHY THE SCOPE IS NARROW (conservatism — the load-bearing design choice).
+// An earlier generalised rule (^<short-prefix>?API Error:\s*\d{3}\b.*$) was TOO
+// BROAD: its unbounded `.*` tail let any short prefix + "API Error: NNN" + an
+// arbitrarily long sentence match, so it KILLED legitimate long answers that merely
+// DISCUSS an API error (e.g. "API Error: 500 happened because the server was
+// overloaded. To fix this, retry with exponential backoff …"). That is the worst
+// outcome: a false-positive costs the user a missing answer AND a double-burn retry,
+// whereas the rare false-negative (caching one transient error for the 5-min TTL) is
+// cheap and self-healing. So C-1 is reframed from "detect ANY API error" to "detect
+// a claude-CLI AUTHENTICATION-FAILURE banner", and when unsure it PASSES (does not
+// kill). Transient 5xx server errors are deliberately NOT detected — they are not the
+// R-1 case and the conservative choice is to let them through.
+//
+// THE SIGNAL — a turn is an auth-failure banner only if ALL of these hold over the
+// WHOLE trimmed assistant text (a conjunction; any one failing => PASS):
+//   1. SHORT whole-message. Real banners are one short line (the two live samples are
+//      69 and 73 chars). Cap = TUI_ERR_MAX_LEN (100) — headroom over 73 for a
+//      slightly longer future banner, while still rejecting multi-sentence prose. A
+//      long answer that happens to discuss auth (no code chars, e.g. 226 chars) is
+//      rejected on length alone.
+//   2. Contains "API Error: 4\d{2}" — auth failures are 4xx (401/403). This rejects
+//      transient 5xx ("API Error: 500/503 …") and bare "HTTP 401 means unauthorized."
+//      (no "API Error:" core).
+//   3. Contains an auth KEYWORD — authenticat | /login | credential (case-insensitive).
+//      This rejects answers that quote a 4xx but are not auth banners, e.g.
+//      "To debug a 401: the server returns API Error: 401 Unauthorized …"
+//      ("Unauthorized" is authoriz-, not authenticat-; no /login, no credential).
+//   4. Contains NO backtick or quote char (` ' "). A real CLI banner is plain text;
+//      backticked/quoted text signals an answer that is QUOTING the error rather than
+//      being the banner, e.g. "You'll see `API Error: 401` … run /login to fix it."
+//      (75 chars — passes 1-3 but is excluded here). This is the conservative tie-
+//      breaker for short instructional answers.
+//
+// Worked matrix (all required cases pass — see test-features.mjs C-1 block):
+//   KILL: "Please run /login · API Error: 401 Invalid authentication credentials"
+//   KILL: "Failed to authenticate. API Error: 401 Invalid authentication credentials"
+//   PASS: "API Error: 500 happened because the server was overloaded. …"      (not 4xx)
+//   PASS: "Failed to parse the config. Here are the API Error: 401 details …"  (too long + no auth-kw)
+//   PASS: "To debug a 401: … API Error: 401 Unauthorized, then you refresh …"  (no auth-kw)
+//   PASS: "Here is the handler … It logs the string API Error: 503 …"          (not 4xx)
+//   PASS: "You'll see `API Error: 401` … run /login to fix it."                (has backtick)
+//   PASS: "HTTP 401 means unauthorized."                                       (no API Error core)
+//   PASS: "The capital of France is Paris."                                    (nothing matches)
+//
+// OPERATOR OVERRIDE (unchanged): CLAUDE_TUI_ERROR_PATTERNS lets an operator REPLACE
+// the default auth-banner detector with their own newline- or `||`-separated JS regex
+// source strings (each auto-anchored ^…$ over the trimmed text, case-insensitive). A
+// non-empty override uses ONLY those regexes (the narrowed default is bypassed); an
+// empty / whitespace-only override DISABLES detection entirely (escape hatch).
+
+// Whole-message length cap for the default auth-banner detector. Real banners are
+// 69/73 chars; 100 gives headroom while still rejecting multi-sentence prose.
+const TUI_ERR_MAX_LEN = 100;
+// 4xx "API Error:" core — auth failures are 4xx (401/403), never 5xx.
+const TUI_ERR_4XX = /API Error:\s*4\d{2}\b/i;
+// Auth keyword — the message must be about authentication, not just quote a 4xx.
+const TUI_ERR_AUTH_KW = /authenticat|\/login|credential/i;
+// Code/quote chars — their presence signals prose QUOTING an error, not the banner.
+const TUI_ERR_CODE_CHAR = /[`'"]/;
+
+// Default detector: returns true iff `trimmed` IS a claude-CLI auth-failure banner
+// (all four signals above). Conservative — any signal failing => false (PASS).
+function isDefaultAuthFailureBanner(trimmed) {
+  if (trimmed.length > TUI_ERR_MAX_LEN) return false; // 1. short whole-message
+  if (!TUI_ERR_4XX.test(trimmed)) return false;       // 2. 4xx API Error core
+  if (!TUI_ERR_AUTH_KW.test(trimmed)) return false;   // 3. auth keyword
+  if (TUI_ERR_CODE_CHAR.test(trimmed)) return false;  // 4. no code/quote chars
+  return true;
+}
+
+// Compile an OPERATOR-SUPPLIED pattern set (override path only). Each source is
+// anchored ^…$ over the trimmed text and matched case-insensitively (`s` so `.` spans
+// a multi-line banner). A pattern that fails to compile is skipped (never throws into
+// the request path).
+function compileTuiErrorPatterns(raw) {
+  const sources = String(raw).split(/\r?\n|\|\|/).map((s) => s.trim()).filter(Boolean);
+  const out = [];
+  for (const src of sources) {
+    try { out.push(new RegExp(`^(?:${src})$`, "is")); } catch { /* skip bad pattern */ }
+  }
+  return out;
+}
+
+// Returns the matched banner text (the trimmed assistant text) if `text` IS a claude-
+// CLI auth-failure banner in its entirety, else null. `patternsRaw` defaults to
+// process.env.CLAUDE_TUI_ERROR_PATTERNS:
+//   - undefined  → narrowed default auth-banner detector (isDefaultAuthFailureBanner).
+//   - non-empty  → operator regex override REPLACES the default.
+//   - empty/ws   → detection disabled (escape hatch).
+export function detectTuiUpstreamError(text, patternsRaw = process.env.CLAUDE_TUI_ERROR_PATTERNS) {
+  if (typeof text !== "string") return null;
+  const trimmed = text.trim();
+  if (!trimmed) return null;
+  if (patternsRaw == null) {
+    return isDefaultAuthFailureBanner(trimmed) ? trimmed : null;
+  }
+  // Operator override path: empty/whitespace disables; otherwise use only their regexes.
+  const patterns = compileTuiErrorPatterns(patternsRaw);
+  if (patterns.length === 0) return null;
+  for (const re of patterns) {
+    if (re.test(trimmed)) return trimmed;
+  }
+  return null;
+}
+
+// Block until the session transcript is terminal (turn_duration / final
+// stop_reason) or the wall-clock cap elapses, polling the file (no fs.watch —
+// robust over NFS / editors). Returns { text, entrypoint, truncated }:
+//   - text:       latest assistant text.
+//   - entrypoint: billing-pool classifier (see verifyEntrypoint), or null.
+//   - truncated:  FALSE when a terminal marker was reached (the turn completed);
+//                 TRUE when the wall-clock cap was hit with partial text but NO
+//                 terminal marker (the turn is INCOMPLETE — what we have is a
+//                 cut-off prefix). (C-2, issue #133.)
+//
+// Why `truncated` matters: previously the terminal-marker path and the
+// cap-with-partial-text path BOTH returned `{text, entrypoint}` identically, so
+// callClaudeTui could not tell a complete answer from a truncated one and cached +
+// returned the partial as finish_reason:stop (silent success). The caller now
+// throws on `truncated` so a cut-off turn is neither cached nor counted as success.
+// The field is additive — existing call sites that ignore it keep working.
+//
+// On cap with NO text at all, still throws (unchanged) — there is nothing to return.
+//
+// No quiescence heuristic by design: a long Opus thinking turn stalls transcript
+// growth and a "file stable for N s" rule would false-abort it (spec §4.3).
+// Resolution: pass an explicit `transcriptPath` (used by unit tests), OR pass
+// `home` + `sessionId` to resolve by glob each poll (production) — the transcript
+// file does not exist until the turn starts, so resolution happens inside the loop.
+// `abortSignal` (optional): when it fires, stop waiting and throw TuiAbortError. The one
+// caller that passes it is the STREAMING TUI path, which ties it to the client's socket:
+// a client that disconnects mid-turn should not leave the pane running (and the caller's
+// concurrency slot held) until the turn or the 120s cap ends. runTuiTurn's finally does the
+// teardown. Omitted => the loop is byte-for-byte the pre-streaming loop.
+export async function readTuiTranscript({ transcriptPath: p, home, sessionId, wallclockMs = 120000, pollMs = 250, abortSignal = null }) {
+  const deadline = Date.now() + wallclockMs;
+  let lastText = "";
+  let lastEntrypoint = null;
+  while (Date.now() < deadline) {
+    if (abortSignal && abortSignal.aborted) {
+      const err = new Error("tui_aborted: client disconnected before the turn completed");
+      err.name = "TuiAbortError";
+      throw err;
+    }
+    const resolved = p || findTranscriptPath(home, sessionId);
+    if (resolved && existsSync(resolved)) {
+      const events = parseTranscriptLines(readFileSync(resolved, "utf8"));
+      lastText = extractLatestAssistantText(events) || lastText;
+      const ep = verifyEntrypoint(events);
+      if (ep != null) lastEntrypoint = ep;
+      // Terminal marker reached → the turn is COMPLETE.
+      if (events.some(isTerminalLine)) return { text: lastText, entrypoint: lastEntrypoint, truncated: false };
+    }
+    await sleep(pollMs);
+  }
+  // Cap elapsed with no terminal marker. If we have partial text, flag it truncated
+  // so the caller rejects it (don't cache / don't count as success). No text at all
+  // → throw (nothing to return).
+  if (lastText) return { text: lastText, entrypoint: lastEntrypoint, truncated: true };
+  throw new Error("tui_transcript_timeout: no assistant text within wallclock cap");
+}

models.json

@@ -2,6 +2,14 @@
   "$schema": "./models.schema.json",
   "version": 1,
   "models": [
+    {
+      "id": "claude-opus-4-8",
+      "displayName": "Claude Opus 4.8",
+      "openclawName": "Claude Opus 4.8 (via CLI)",
+      "reasoning": true,
+      "contextWindow": 200000,
+      "maxTokens": 16384
+    },
     {
       "id": "claude-opus-4-7",
       "displayName": "Claude Opus 4.7",
@@ -18,6 +26,14 @@
       "contextWindow": 200000,
       "maxTokens": 16384
     },
+    {
+      "id": "claude-sonnet-5",
+      "displayName": "Claude Sonnet 5",
+      "openclawName": "Claude Sonnet 5 (via CLI)",
+      "reasoning": true,
+      "contextWindow": 200000,
+      "maxTokens": 16384
+    },
     {
       "id": "claude-sonnet-4-6",
       "displayName": "Claude Sonnet 4.6",
@@ -36,8 +52,8 @@
     }
   ],
   "aliases": {
-    "opus": "claude-opus-4-7",
-    "sonnet": "claude-sonnet-4-6",
+    "opus": "claude-opus-4-8",
+    "sonnet": "claude-sonnet-5",
     "haiku": "claude-haiku-4-5-20251001"
   },
   "legacyAliases": {

ocp

@@ -573,21 +573,42 @@ Usage:
   ocp restart            Restart the Claude proxy service
   ocp restart gateway    Restart the OpenClaw gateway
                          (briefly disconnects all Telegram/Discord bots)
+
+Note (macOS): restart does a full launchctl bootout + bootstrap, NOT
+`kickstart -k`. bootout+bootstrap re-reads the plist's EnvironmentVariables,
+so an env change you made (e.g. CLAUDE_BIND, CLAUDE_CODE_OAUTH_TOKEN) actually
+takes effect. `kickstart -k` only re-execs the process and reuses launchd's
+cached env, so env edits would be silently ignored. (Linux systemctl already
+re-reads its EnvironmentFile on restart.)
 EOF
 }
 
+# macOS only: reload a launchd agent via bootout + bootstrap so plist
+# EnvironmentVariables are re-read (kickstart -k would reuse the cached env).
+# Args: <uid> <label> <plist-path>. Returns 0 iff bootstrap succeeds.
+_launchd_reload() {
+  local uid="$1" label="$2" plist="$3"
+  [[ -f "$plist" ]] || return 1
+  # bootout may legitimately fail if the agent is not currently loaded — that's fine,
+  # we only require the subsequent bootstrap to succeed (the load that re-reads env).
+  launchctl bootout "gui/$uid/$label" 2>/dev/null || true
+  launchctl bootstrap "gui/$uid" "$plist" 2>/dev/null
+}
+
 cmd_restart() {
   if [[ "${1:-}" == "gateway" ]]; then
     echo "Restarting gateway..."
     openclaw gateway restart 2>&1
   else
     echo "Restarting proxy..."
-    # Try current service name, then legacy, then manual restart
+    # Try current service name, then legacy, then manual restart.
+    # macOS: bootout+bootstrap (re-reads plist EnvironmentVariables — see cmd_restart_help).
+    # Linux: systemctl --user restart already re-reads its EnvironmentFile.
     local uid
     uid=$(id -u)
-    if launchctl kickstart -k "gui/$uid/dev.ocp.proxy" 2>/dev/null; then
+    if _launchd_reload "$uid" "dev.ocp.proxy" "$HOME/Library/LaunchAgents/dev.ocp.proxy.plist"; then
       true
-    elif launchctl kickstart -k "gui/$uid/ai.openclaw.proxy" 2>/dev/null; then
+    elif _launchd_reload "$uid" "ai.openclaw.proxy" "$HOME/Library/LaunchAgents/ai.openclaw.proxy.plist"; then
       true
     elif systemctl --user restart ocp-proxy 2>/dev/null; then
       true
@@ -601,7 +622,12 @@ cmd_restart() {
       self_r="${BASH_SOURCE[0]}"
       while [[ -L "$self_r" ]]; do self_r="$(readlink "$self_r")"; done
       script_dir="$(cd "$(dirname "$self_r")" && pwd)"
-      DISABLE_AUTOUPDATER=1 nohup node "$script_dir/server.mjs" >> "$HOME/.ocp/logs/proxy.log" 2>&1 &
+      # env -u strips test-only key-store redirection vars (A4): if the invoking shell had
+      # NODE_ENV=test + OCP_DIR_OVERRIDE exported (e.g. from a debugging session), this manual
+      # fallback would otherwise inherit them and start the daemon against a scratch/empty key
+      # store — a silent auth outage in AUTH_MODE=multi. The plist/systemd paths strip these via
+      # plist-merge's NEVER_PRESERVE; this covers the one direct-launch path OCP controls.
+      DISABLE_AUTOUPDATER=1 env -u NODE_ENV -u OCP_DIR_OVERRIDE nohup node "$script_dir/server.mjs" >> "$HOME/.ocp/logs/proxy.log" 2>&1 &
     fi
     sleep 3
     if curl -sf --max-time 5 "$PROXY/health" > /dev/null 2>&1; then
@@ -709,6 +735,8 @@ Usage:
   ocp update --rollback --list        List available snapshots
   ocp update --rollback <path>        Restore a specific snapshot
   ocp update --rollback --dry-run     Preview rollback plan
+  ocp update --rollback --gc          Delete old snapshots (keep last 5, or <30 days)
+  ocp update --rollback --gc --dry-run    Preview what would be deleted
 EOF
 }
 

ocp-connect

@@ -122,11 +122,17 @@ provider = {
     "models": []
 }
 
-# Model metadata mapping (prefix match for versioned IDs like claude-haiku-4-5-20251001)
+# Model metadata mapping. Prefix match on the model FAMILY (claude-opus / -sonnet /
+# -haiku), not a pinned version. A version-pinned prefix like "claude-sonnet-4"
+# silently misses "claude-sonnet-5" and falls through to the non-reasoning /
+# 8k-output default (PR #152 review) — every future Sonnet/Opus/Haiku bump would
+# re-trip it. Family prefixes classify any versioned ID correctly with no per-model
+# edit. (ADR 0003: models.json is the SPOT for model existence; /v1/models does not
+# expose reasoning/maxTokens, so family classification stays here.)
 model_meta = {
-    "claude-opus-4": {"name": "Claude Opus (OCP)", "reasoning": True, "maxTokens": 16384},
-    "claude-sonnet-4": {"name": "Claude Sonnet (OCP)", "reasoning": True, "maxTokens": 16384},
-    "claude-haiku-4": {"name": "Claude Haiku (OCP)", "reasoning": False, "maxTokens": 8192},
+    "claude-opus": {"name": "Claude Opus (OCP)", "reasoning": True, "maxTokens": 16384},
+    "claude-sonnet": {"name": "Claude Sonnet (OCP)", "reasoning": True, "maxTokens": 16384},
+    "claude-haiku": {"name": "Claude Haiku (OCP)", "reasoning": False, "maxTokens": 8192},
 }
 
 def get_model_meta(mid):
@@ -178,11 +184,11 @@ config.setdefault("agents", {})
 config["agents"].setdefault("defaults", {})
 config["agents"]["defaults"].setdefault("models", {})
 
-# Build alias map (prefix match)
+# Build alias map (family prefix match — version-agnostic, see model_meta note)
 alias_prefixes = {
-    "claude-opus-4": "Claude Opus",
-    "claude-sonnet-4": "Claude Sonnet",
-    "claude-haiku-4": "Claude Haiku",
+    "claude-opus": "Claude Opus",
+    "claude-sonnet": "Claude Sonnet",
+    "claude-haiku": "Claude Haiku",
 }
 
 for mid in model_ids:
@@ -196,8 +202,13 @@ for mid in model_ids:
 
 # Handle primary/backup
 if priority == "1":
-    # OCP as primary — pick the best model (prefer sonnet for daily use)
-    primary_model = provider_name + "/claude-sonnet-4-6" if "claude-sonnet-4-6" in model_ids else provider_name + "/" + model_ids[0]
+    # OCP as primary — pick the best model (prefer the latest Sonnet for daily use,
+    # tracking the `sonnet` alias default in models.json; fall back across versions).
+    _sonnet_pref = ["claude-sonnet-5", "claude-sonnet-4-6"]
+    primary_model = next(
+        (provider_name + "/" + m for m in _sonnet_pref if m in model_ids),
+        provider_name + "/" + model_ids[0],
+    )
     config["agents"]["defaults"].setdefault("model", {})
     config["agents"]["defaults"]["model"]["primary"] = primary_model
     # Keep existing fallbacks
@@ -506,9 +517,11 @@ main() {
   echo ""
 
   # Step 2.5: auto-discover anonymous key from /health (issue #12 §14 Path A).
-  # When the OCP admin set PROXY_ANONYMOUS_KEY, the server advertises it via
-  # /health.anonymousKey. If the user didn't pass --key, use it automatically so
-  # `ocp-connect <host>` works zero-config for OpenClaw multi-agent setups.
+  # The server advertises anonymousKey in /health ONLY when the admin has set
+  # PROXY_ADVERTISE_ANON_KEY=1 (default off — /health is unauthenticated, so
+  # advertising exposes the shared key to any LAN-reachable device; issue #109).
+  # Localhost callers always receive it regardless. When the field is absent,
+  # ocp-connect falls back to anonymous access / interactive --key (step 3 below).
   if [[ -z "$key" ]]; then
     local anon_key
     anon_key=$(echo "$health_json" | python3 -c "
@@ -632,11 +645,12 @@ PYEOF
     {
       echo ""
       echo "# OCP LAN (added by ocp connect)"
-      echo "export OPENAI_BASE_URL=$base_url/v1"
+      echo "export OPENAI_BASE_URL='$base_url/v1'"
       if [[ -n "$key" ]]; then
-        echo "export OPENAI_API_KEY=$key"
+        echo "export OPENAI_API_KEY='$key'"
       fi
     } >> "$rc_file"
+    chmod 600 "$rc_file" 2>/dev/null || true
   done
 
   echo "  Shell config:"
@@ -669,6 +683,7 @@ PYEOF
         echo "OPENAI_API_KEY=$key"
       fi
     } > "$env_dir/ocp.conf"
+    chmod 600 "$env_dir/ocp.conf" 2>/dev/null || true
     echo ""
     echo "  System-level (systemd):"
     echo "    ✓ $env_dir/ocp.conf"

ocp-plugin/index.js

@@ -1,9 +1,19 @@
 /**
  * OCP Plugin — registers /ocp as a native slash command in OpenClaw gateway.
- * Calls the local claude-proxy at http://127.0.0.1:3456 and formats the response.
+ * Calls the local claude-proxy and formats the response.
+ *
+ * Port resolution (in priority order):
+ *   1. OCP_PROXY_URL env (full URL, e.g. http://10.0.0.5:3456)
+ *   2. CLAUDE_PROXY_PORT env (port only; localhost assumed)
+ *   3. Fallback: http://127.0.0.1:3456 (OCP server source default since v1.0)
+ *
+ * If a particular host's OCP plist injects a non-default CLAUDE_PROXY_PORT,
+ * the OpenClaw launchd plist for that host must also inject the same
+ * CLAUDE_PROXY_PORT into the plugin's env, or the plugin will fall back to
+ * 3456 and miss the server.
  */
-
-const PROXY = "http://127.0.0.1:3456";
+const PROXY = process.env.OCP_PROXY_URL
+  || (process.env.CLAUDE_PROXY_PORT ? `http://127.0.0.1:${process.env.CLAUDE_PROXY_PORT}` : "http://127.0.0.1:3456");
 
 // Wrap output in monospace code block for Telegram/Discord alignment
 function mono(text) { return "```\n" + text + "\n```"; }
@@ -198,31 +208,34 @@ async function cmdTest() {
 async function cmdRestart(args) {
   const target = (args || "").trim().toLowerCase();
   const { execSync } = await import("node:child_process");
+  const uid = typeof process.getuid === "function" ? process.getuid() : 501;
+  const macProxy = `launchctl kickstart -k gui/${uid}/dev.ocp.proxy`;
+  const macGateway = `launchctl kickstart -k gui/${uid}/ai.openclaw.gateway`;
   try {
     if (target === "gateway") {
-      execSync("launchctl kickstart -k gui/501/ai.openclaw.gateway", { timeout: 15000 });
+      execSync(macGateway, { timeout: 15000 });
       return "✓ Gateway restarted";
     } else if (target === "all") {
-      execSync("launchctl kickstart -k gui/501/ai.openclaw.proxy", { timeout: 15000 });
+      execSync(macProxy, { timeout: 15000 });
       // Gateway restart will kill this plugin too, so do it last
-      execSync("launchctl kickstart -k gui/501/ai.openclaw.gateway", { timeout: 15000 });
+      execSync(macGateway, { timeout: 15000 });
       return "✓ Proxy + Gateway restarted";
     } else {
-      execSync("launchctl kickstart -k gui/501/ai.openclaw.proxy", { timeout: 15000 });
+      execSync(macProxy, { timeout: 15000 });
       return "✓ Proxy restarted";
     }
   } catch (e) {
-    // Try systemd for Linux
+    // Linux: systemd user services
     try {
       if (target === "gateway") {
         execSync("systemctl --user restart openclaw-gateway", { timeout: 15000 });
         return "✓ Gateway restarted";
       } else {
-        execSync("systemctl --user restart openclaw-proxy 2>/dev/null || pkill -f 'node.*server.mjs' && sleep 2 && cd ~/.openclaw/projects/*/; node server.mjs &", { timeout: 15000, shell: true });
+        execSync("systemctl --user restart ocp-proxy", { timeout: 15000 });
         return "✓ Proxy restarted";
       }
     } catch (e2) {
-      return `✗ Restart failed: ${e2.message?.slice(0, 100)}`;
+      return `✗ Restart failed: ${e2.message?.slice(0, 100)}. Run \`ocp restart\` on the server host manually.`;
     }
   }
 }

ocp-plugin/openclaw.plugin.json

@@ -2,7 +2,7 @@
   "id": "ocp",
   "name": "OCP Commands",
   "description": "Slash commands for the OpenClaw Proxy — /ocp usage, /ocp settings, /ocp health, etc.",
-  "version": "3.12.0",
+  "version": "3.16.2",
   "configSchema": {
     "type": "object",
     "additionalProperties": false,
@@ -10,7 +10,7 @@
       "proxyUrl": {
         "type": "string",
         "default": "http://127.0.0.1:3456",
-        "description": "URL of the Claude proxy"
+        "description": "URL of the Claude proxy. Overridable via OCP_PROXY_URL or CLAUDE_PROXY_PORT env."
       }
     }
   }

ocp-plugin/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ocp",
-  "version": "3.12.0",
+  "version": "3.16.2",
   "description": "Slash commands for the OpenClaw Proxy",
   "main": "index.js",
   "type": "module",
@@ -9,6 +9,7 @@
   "openclaw": {
     "type": "plugin",
     "id": "ocp",
-    "pluginManifest": "openclaw.plugin.json"
+    "pluginManifest": "openclaw.plugin.json",
+    "extensions": ["./index.js"]
   }
 }

package.json

@@ -1,6 +1,6 @@
 {
   "name": "open-claude-proxy",
-  "version": "3.15.0",
+  "version": "3.24.0",
   "description": "OCP (Open Claude Proxy) — use your Claude Pro/Max subscription as an OpenAI-compatible API for any IDE. Works with Cline, OpenCode, Aider, Continue.dev, OpenClaw, and more.",
   "type": "module",
   "bin": {

scripts/doctor.mjs

@@ -15,6 +15,7 @@ import { readFileSync, existsSync } from "node:fs";
 import { join } from "node:path";
 import { homedir } from "node:os";
 import { execSync } from "node:child_process";
+import { DEFAULT_PORT } from "../lib/constants.mjs";
 
 const SCHEMA_VERSION = "1";
 
@@ -37,6 +38,11 @@ export async function runDoctor(opts = {}) {
   const push = (id, level, message, extra = {}) =>
     checks.push({ id, level, message, ...extra });
 
+  // --- fast path: --check oauth ---
+  if (opts.checkOnly === "oauth") {
+    return runOauthOnly(opts, checks, push);
+  }
+
   // --- version detection ---
   const ocpDir = opts.ocpDir || join(homedir(), "ocp");
   let currentVersion = opts.mockVersion;
@@ -48,7 +54,29 @@ export async function runDoctor(opts = {}) {
       currentVersion = "unknown";
     }
   }
-  const latestVersion = opts.mockLatest || "v3.14.0";
+  // Resolve latest from origin/main (cheap: `git show origin/main:package.json`).
+  // Falls back to current_version when network/git unavailable, so kind = noop instead
+  // of recommending a downgrade against a stale hardcoded value.
+  let latestVersion = opts.mockLatest;
+  if (!latestVersion) {
+    // Issue #173: `git show origin/main:...` reads the LOCALLY CACHED remote ref. Without a
+    // fetch first, a machine that hasn't pulled since the last release sees latest == current
+    // and reports noop — new releases were invisible everywhere except the machine that cut
+    // the tag (live repro: Oracle VM, 2026-07-17). Fetch before comparing; on failure
+    // (offline, auth, timeout) fall through to the cached ref — the pre-existing behavior.
+    if (!opts.skipNetwork) {
+      try {
+        execSync(`git -C ${ocpDir} fetch --tags --quiet`, { stdio: ["pipe", "pipe", "pipe"], timeout: 15000 });
+      } catch { /* offline → compare against cached origin/main, as before */ }
+    }
+    try {
+      const out = execSync(`git -C ${ocpDir} show origin/main:package.json 2>/dev/null`, { stdio: ["pipe", "pipe", "pipe"] }).toString();
+      const remotePkg = JSON.parse(out);
+      latestVersion = `v${remotePkg.version}`;
+    } catch {
+      latestVersion = currentVersion;
+    }
+  }
   push("current_version", "PASS", `current=${currentVersion}`);
 
   // --- from-version supported? ---
@@ -64,7 +92,7 @@ export async function runDoctor(opts = {}) {
       health = opts.mockHealth;
     } else {
       try {
-        const port = process.env.CLAUDE_PROXY_PORT || "3478";
+        const port = process.env.CLAUDE_PROXY_PORT || String(DEFAULT_PORT);
         const out = execSync(`curl -sf --max-time 3 http://127.0.0.1:${port}/health`, { stdio: ["pipe", "pipe", "pipe"] }).toString();
         health = { status: 200, body: JSON.parse(out) };
       } catch (e) {
@@ -178,6 +206,84 @@ export async function runDoctor(opts = {}) {
   };
 }
 
+function runOauthOnly(opts, checks, push) {
+  let healthOk = true, oauthOk = true;
+  let health;
+  if (opts.mockHealth !== undefined) {
+    health = opts.mockHealth;
+  } else {
+    try {
+      const port = process.env.CLAUDE_PROXY_PORT || String(DEFAULT_PORT);
+      const out = execSync(`curl -sf --max-time 3 http://127.0.0.1:${port}/health`, { stdio: ["pipe", "pipe", "pipe"] }).toString();
+      health = { status: 200, body: JSON.parse(out) };
+    } catch (e) {
+      health = { error: String(e.message || e) };
+    }
+  }
+
+  if (health.error || health.status !== 200) {
+    healthOk = false;
+    push("oauth_ok", "FAIL", `service unreachable: ${health.error || `status ${health.status}`}`);
+  } else if (!health.body || typeof health.body !== "object") {
+    healthOk = false;
+    push("oauth_ok", "FAIL", "service /health returned 200 but empty/non-JSON body");
+  } else if (!health.body?.auth?.ok) {
+    oauthOk = false;
+    push("oauth_ok", "FAIL", `auth.ok=false: ${health.body?.auth?.message || "unknown"}`);
+  } else {
+    push("oauth_ok", "PASS", "OAuth token valid");
+  }
+
+  const kind = !healthOk ? "fix_service" : !oauthOk ? "fix_oauth" : "noop";
+
+  let next_action;
+  const ocpDir = opts.ocpDir || join(homedir(), "ocp");
+  if (kind === "noop") {
+    next_action = { kind, human_required: [], ai_executable: [], verify: "OAuth healthy" };
+  } else if (kind === "fix_oauth") {
+    next_action = {
+      kind,
+      human_required: [],
+      ai_executable: [
+        `cd "$(npm root -g)/@anthropic-ai/claude-code" && node install.cjs`,
+        `launchctl bootout gui/$(id -u)/dev.ocp.proxy 2>/dev/null || true`,
+        `launchctl bootstrap gui/$(id -u) ${join(homedir(), "Library", "LaunchAgents", "dev.ocp.proxy.plist")}`,
+        `${ocpDir}/ocp doctor --check oauth`
+      ],
+      verify: "ocp doctor --check oauth expects PASS",
+      reference: "~/.cc-rules/memory/learnings/ocp_claude_native_binary_postinstall.md"
+    };
+  } else {
+    next_action = {
+      kind,
+      human_required: [],
+      ai_executable: [
+        `launchctl bootout gui/$(id -u)/dev.ocp.proxy 2>/dev/null || true`,
+        `launchctl bootstrap gui/$(id -u) ${join(homedir(), "Library", "LaunchAgents", "dev.ocp.proxy.plist")}`,
+        `${ocpDir}/ocp doctor --check oauth`
+      ],
+      verify: "ocp doctor --check oauth expects service_running=PASS"
+    };
+  }
+
+  const fail_count = checks.filter(c => c.level === "FAIL").length;
+  // "skipped" = --check oauth fast path intentionally omits version detection.
+  // AI agents should NOT semver-compare against current_version/latest_version when
+  // either equals "skipped"; the full path provides those fields when needed.
+  return {
+    schema_version: SCHEMA_VERSION,
+    timestamp: new Date().toISOString(),
+    ready_to_upgrade: fail_count === 0,
+    current_version: opts.mockVersion || "skipped",
+    latest_version: opts.mockLatest || "skipped",
+    from_version_supported: true,
+    fail_count,
+    warn_count: 0,
+    checks,
+    next_action
+  };
+}
+
 // CLI entrypoint — use fileURLToPath + realpath to handle symlinked install paths
 // (e.g. /tmp/ → /private/tmp/ on macOS would otherwise miss the guard).
 import { fileURLToPath } from "node:url";
@@ -190,7 +296,9 @@ function _isMain() {
 }
 if (_isMain()) {
   const wantJson = process.argv.includes("--json");
-  const result = await runDoctor();
+  const checkIdx = process.argv.indexOf("--check");
+  const checkOnly = checkIdx !== -1 ? process.argv[checkIdx + 1] : undefined;
+  const result = await runDoctor({ checkOnly });
   if (wantJson) {
     console.log(JSON.stringify(result, null, 2));
   } else {

scripts/lib/plist-merge.mjs

@@ -8,7 +8,22 @@
 //
 // No new dependencies — regex-based, plist <key>X</key><string>Y</string> shape
 // is stable enough for our hand-written templates in setup.mjs.
+//
+// SECURITY DENYLIST (A4): keys that must NEVER be carried into a service unit, even when a
+// prior unit already contained them. OCP's key store honors OCP_DIR_OVERRIDE only when
+// NODE_ENV === "test" (keys.mjs). If BOTH somehow reached a daemon's environment, the server
+// would open a scratch/empty key store instead of ~/.ocp/ocp.db — in AUTH_MODE=multi a silent
+// total auth outage. The preservation rule below ("keys only in EXISTING are kept verbatim")
+// is exactly a vector for that: a unit that once carried these test-only vars would otherwise
+// survive every setup re-run. So we strip them from the preserved set unconditionally. This is
+// defense-in-depth: setup.mjs's own template never injects them, so the only way they enter is
+// preservation, and this closes it. (The residual path — a hand-rolled `node server.mjs` with
+// both vars exported — is out of any launcher's reach; keys.mjs's loud "NOT the default" log is
+// the backstop there.)
+export const NEVER_PRESERVE = new Set(["NODE_ENV", "OCP_DIR_OVERRIDE"]);
 
+// Note: setup.mjs XML-escapes all injected values before writing (via xmlEscape()),
+// so raw `<` / `>` / `&` never appear in plist <string> bodies — the [^<]* regex below is safe.
 const PLIST_KV_RE = /<key>([^<]+)<\/key>\s*<string>([^<]*)<\/string>/g;
 
 export function parsePlistEnv(plistContent) {
@@ -34,7 +49,7 @@ export function mergePlistEnv(existing, template) {
 
   const preserved = {};
   for (const [k, v] of Object.entries(existingEnv)) {
-    if (!KNOWN.has(k)) preserved[k] = v;
+    if (!KNOWN.has(k) && !NEVER_PRESERVE.has(k)) preserved[k] = v;
   }
   if (Object.keys(preserved).length === 0) return template;
 
@@ -70,7 +85,7 @@ export function mergeSystemdEnv(existing, template) {
   const KNOWN = new Set(Object.keys(templateEnv));
 
   const preservedLines = Object.entries(existingEnv)
-    .filter(([k]) => !KNOWN.has(k))
+    .filter(([k]) => !KNOWN.has(k) && !NEVER_PRESERVE.has(k))
     .map(([k, v]) => `Environment=${k}=${v}`);
   if (preservedLines.length === 0) return template;
 

scripts/lib/snapshot.mjs

@@ -1,8 +1,8 @@
-import { mkdirSync, writeFileSync, readFileSync, copyFileSync, existsSync, readdirSync, statSync } from "node:fs";
+import { mkdirSync, writeFileSync, readFileSync, copyFileSync, existsSync, readdirSync, statSync, rmSync } from "node:fs";
 import { join } from "node:path";
 
 export function writeSnapshot({ homeDir, fromCommit, fromVersion, toVersion, extraFiles = [] }) {
-  const ts = new Date().toISOString().replace(/\.\d+Z$/, "Z");
+  const ts = formatSnapshotTimestamp(new Date());
   const root = join(homeDir, ".ocp", `upgrade-snapshot-${ts}`);
   mkdirSync(root, { recursive: true });
 
@@ -48,5 +48,83 @@ export function listSnapshots(homeDir) {
   return readdirSync(root)
     .filter(name => name.startsWith("upgrade-snapshot-"))
     .map(name => ({ name, path: join(root, name), mtime: statSync(join(root, name)).mtimeMs }))
-    .sort((a, b) => a.name.localeCompare(b.name));
+    .sort((a, b) => {
+      const chronological = parseSnapshotTimestamp(a.name) - parseSnapshotTimestamp(b.name);
+      return chronological || a.name.localeCompare(b.name);
+    });
+}
+
+/**
+ * Garbage-collect old upgrade snapshots.
+ *
+ * Retention rule (a snapshot is KEPT if any of these is true):
+ *   - It is among the last `keepCount` snapshots (sorted oldest→newest)
+ *   - Its timestamp is within `keepDays` of `now`
+ *   - It is the single most-recent snapshot (always-keep safety net)
+ *
+ * @param {string} homeDir - Root containing ~/.ocp/
+ * @param {object} opts
+ * @param {number} [opts.keepCount=5] - Minimum count to keep
+ * @param {number} [opts.keepDays=30] - Keep snapshots newer than N days
+ * @param {boolean} [opts.dryRun=false] - If true, report plan but don't delete
+ * @param {Date} [opts.now=new Date()] - Override clock for testing
+ * @returns {{kept: Array, removed: Array, dryRun: boolean}}
+ */
+export function gcSnapshots(homeDir, opts = {}) {
+  const keepCount = opts.keepCount ?? 5;
+  const keepDays = opts.keepDays ?? 30;
+  const dryRun = !!opts.dryRun;
+  const now = opts.now || new Date();
+
+  const all = listSnapshots(homeDir);  // sorted oldest→newest
+  if (all.length === 0) return { kept: [], removed: [], dryRun };
+  if (all.length === 1) return { kept: all, removed: [], dryRun };  // always keep most recent
+
+  const cutoffMs = now.getTime() - keepDays * 24 * 60 * 60 * 1000;
+  const lastN = new Set(all.slice(-keepCount).map(s => s.path));
+
+  const kept = [], removed = [];
+  for (let i = 0; i < all.length; i++) {
+    const s = all[i];
+    const isMostRecent = i === all.length - 1;
+    const isInLastN = lastN.has(s.path);
+    const isWithinDays = parseSnapshotTimestamp(s.name) >= cutoffMs;
+    if (isMostRecent || isInLastN || isWithinDays) {
+      kept.push(s);
+    } else {
+      removed.push(s);
+    }
+  }
+
+  if (!dryRun) {
+    for (const s of removed) {
+      try {
+        rmSync(s.path, { recursive: true, force: true });
+      } catch (err) {
+        console.error(`[snapshot] warn: could not remove ${s.path} (${err.code || err.message})`);
+      }
+    }
+  }
+
+  return { kept, removed, dryRun };
+}
+
+function parseSnapshotTimestamp(name) {
+  // Both legacy ISO names and Windows-safe names are supported.
+  // upgrade-snapshot-2026-05-11T08:30:00Z → epoch ms
+  // upgrade-snapshot-2026-05-11T08-30-00Z → epoch ms
+  const m = name.match(/upgrade-snapshot-(.+)$/);
+  if (!m) return 0;
+  const raw = m[1];
+  const t = Date.parse(raw);
+  if (Number.isFinite(t)) return t;
+  const iso = raw.replace(/(T\d{2})-(\d{2})-(\d{2})Z$/, "$1:$2:$3Z");
+  const portable = Date.parse(iso);
+  return Number.isFinite(portable) ? portable : 0;
+}
+
+function formatSnapshotTimestamp(date) {
+  // Windows forbids ':' in directory names. Replacing only the time separators
+  // preserves chronological lexical order and keeps the timestamp readable.
+  return date.toISOString().replace(/\.\d+Z$/, "Z").replace(/:/g, "-");
 }

scripts/sync-openclaw.mjs

@@ -9,6 +9,7 @@ import { readFileSync, writeFileSync, existsSync, copyFileSync } from "node:fs";
 import { join, dirname } from "node:path";
 import { fileURLToPath } from "node:url";
 import { homedir } from "node:os";
+import { DEFAULT_PORT, LOCAL_HOST, OPENAI_API_BASE } from "../lib/constants.mjs";
 
 const __dirname = dirname(fileURLToPath(import.meta.url));
 const REPO_ROOT = join(__dirname, "..");
@@ -70,7 +71,7 @@ if (!config.models.providers) config.models.providers = {};
 if (!config.models.providers[PROVIDER_NAME]) {
   // First-time registration
   config.models.providers[PROVIDER_NAME] = {
-    baseUrl: "http://127.0.0.1:3456/v1",
+    baseUrl: `http://${LOCAL_HOST}:${DEFAULT_PORT}${OPENAI_API_BASE}`,
     api: "openai-completions",
     authHeader: false,
     models: desiredModels,

scripts/upgrade.mjs

@@ -14,7 +14,22 @@ import { execSync } from "node:child_process";
 import { homedir } from "node:os";
 import { join } from "node:path";
 import { existsSync, copyFileSync } from "node:fs";
-import { writeSnapshot, listSnapshots, readSnapshot } from "./lib/snapshot.mjs";
+import { writeSnapshot, listSnapshots, readSnapshot, gcSnapshots } from "./lib/snapshot.mjs";
+import { DEFAULT_PORT } from "../lib/constants.mjs";
+
+// Post-flight acceptance predicate (issue #173). A health probe passes ONLY when the server
+// is authed AND actually serving the TARGET version. auth.ok alone is not enough: a stale
+// process holding the port answers auth.ok=true while still running the OLD code — exactly
+// what a nohup-fallback orphan did on 2026-07-17 (upgrade "succeeded", /health kept serving
+// 3.21.1). Comparing /health.version to the checkout target catches orphan-holds-port,
+// restart-didn't-take, and wrong-unit-restarted alike. `target` tolerates a leading "v"
+// (doctor reports "v3.22.1"; /health reports "3.22.1"); an empty/unknown target degrades to
+// the old auth-only check rather than blocking an otherwise-good upgrade.
+export function postFlightOk(body, target) {
+  if (body?.auth?.ok !== true) return false;
+  const want = String(target || "").replace(/^v/, "");
+  return !want || body?.version === want;
+}
 
 export async function runUpgrade(opts = {}) {
   const dryRun = !!opts.dryRun;
@@ -134,18 +149,24 @@ async function runFullUpgrade({ doctor, opts }) {
 
     // phase 6: post-flight (10s budget; skipped under mockExec)
     if (!opts.mockExec) {
-      const port = process.env.CLAUDE_PROXY_PORT || "3478";
+      const port = process.env.CLAUDE_PROXY_PORT || String(DEFAULT_PORT);
       let ok = false;
+      let lastSeen = null;
       for (let i = 0; i < 10; i++) {
         try {
           const out = execSync(`curl -sf --max-time 2 http://127.0.0.1:${port}/health`).toString();
           const body = JSON.parse(out);
-          if (body.auth?.ok === true) { ok = true; break; }
+          lastSeen = body.version;
+          if (postFlightOk(body, doctor.latest_version)) { ok = true; break; }
         } catch { /* retry */ }
         await new Promise(r => setTimeout(r, 1000));
       }
       if (!ok) {
-        phases.push({ name: "post-flight", status: "fail", message: "health did not return auth.ok=true within 10s" });
+        phases.push({
+          name: "post-flight", status: "fail",
+          message: `health did not return auth.ok=true AND version=${doctor.latest_version} within 10s`
+            + (lastSeen ? ` (last saw version=${lastSeen} — a stale process may still hold the port; check \`ss -ltnp\` / \`lsof -i\`)` : ""),
+        });
         throw new Error("post-flight failed");
       }
       execSync(`curl -sf --max-time 3 http://127.0.0.1:${port}/v1/models > /dev/null`);
@@ -154,6 +175,16 @@ async function runFullUpgrade({ doctor, opts }) {
       phases.push({ name: "post-flight", status: "skipped-mock" });
     }
 
+    // Auto-GC old snapshots after successful upgrade (best-effort, never throws).
+    try {
+      const gc = gcSnapshots(homedir(), { keepCount: 5, keepDays: 30 });
+      if (gc.removed.length > 0) {
+        console.error(`[gc] removed ${gc.removed.length} old snapshots; kept ${gc.kept.length}`);
+      }
+    } catch (e) {
+      console.error(`[gc] warn: snapshot GC failed: ${e.message}`);
+    }
+
     return { path: "upgrade", executed: true, changed: true, snapshotPath, phases };
   } catch (err) {
     if (snapshotPath && !err.snapshotPath) {
@@ -193,6 +224,11 @@ async function runRollback(opts) {
   const homeDir = opts.homeDir || homedir();
   const snapshots = opts.mockSnapshots ?? listSnapshots(homeDir);
 
+  if (opts.gc) {
+    const result = gcSnapshots(homeDir, { dryRun: opts.dryRun });
+    return { path: opts.dryRun ? "rollback-gc-dry-run" : "rollback-gc", ...result };
+  }
+
   if (opts.list) {
     return { path: "rollback-list", snapshots };
   }
@@ -295,6 +331,7 @@ if (_isMain()) {
   const yes = args.includes("--yes");
   const rollback = args.includes("--rollback");
   const list = args.includes("--list");
+  const gc = args.includes("--gc");
   const targetIdx = args.indexOf("--target");
   const target = targetIdx !== -1 ? args[targetIdx + 1] : undefined;
   // First non-flag positional after --rollback is the snapshot path
@@ -305,7 +342,7 @@ if (_isMain()) {
     if (cand && !cand.startsWith("--")) snapshotPath = cand;
   }
   try {
-    const result = await runUpgrade({ dryRun, yes, rollback, list, snapshotPath, target });
+    const result = await runUpgrade({ dryRun, yes, rollback, list, gc, snapshotPath, target });
     if (result.plan) for (const line of result.plan) console.log(line);
     if (result.phases) for (const p of result.phases) console.log(`[${p.name}] ${p.status}${p.cmd ? `: ${p.cmd}` : ""}`);
     if (result.steps) for (const s of result.steps) console.log(`  ${s.status === "ok" ? "✓" : s.status === "skipped-mock" ? "·" : "✗"} ${s.cmd}`);
@@ -313,6 +350,10 @@ if (_isMain()) {
       console.log(`Found ${result.snapshots.length} snapshots:`);
       for (const s of result.snapshots) console.log(`  ${s.name}`);
     }
+    if (result.removed && result.kept) {
+      console.log(`Snapshots: kept ${result.kept.length}, ${result.dryRun ? "would remove" : "removed"} ${result.removed.length}`);
+      for (const s of result.removed) console.log(`  - ${s.name}`);
+    }
     process.exit(0);
   } catch (e) {
     console.error(`✗ ${e.message}`);

setup.mjs

@@ -3,7 +3,8 @@
  * OCP (Open Claude Proxy) setup
  *
  * Automatically configures OpenClaw to use Claude CLI as a model provider.
- * Run: node setup.mjs [--port 3456] [--default-model opus|sonnet|haiku] [--dry-run]
+ * Run: node setup.mjs [--port N] [--default-model opus|sonnet|haiku] [--dry-run]
+ *      (default port = DEFAULT_PORT from lib/constants.mjs)
  *
  * What it does:
  *   1. Verifies claude CLI is installed and authenticated
@@ -18,6 +19,7 @@ import { execSync } from "node:child_process";
 import { join, dirname } from "node:path";
 import { homedir } from "node:os";
 import { fileURLToPath } from "node:url";
+import { DEFAULT_PORT } from "./lib/constants.mjs";
 
 const __dirname = dirname(fileURLToPath(import.meta.url));
 const HOME = homedir();
@@ -32,7 +34,7 @@ const opt = (name, fallback) => {
   return i >= 0 && args[i + 1] ? args[i + 1] : fallback;
 };
 
-const PORT = parseInt(opt("port", "3456"), 10);
+const PORT = parseInt(opt("port", String(DEFAULT_PORT)), 10);
 const DEFAULT_MODEL = opt("default-model", "opus"); // opus | sonnet | haiku
 const DRY_RUN = flag("dry-run");
 const SKIP_START = flag("no-start");
@@ -63,6 +65,28 @@ const OCP_ADMIN_KEY_INJECT = process.env.OCP_ADMIN_KEY || null;
 // PROXY_ANONYMOUS_KEY — same pattern
 const PROXY_ANON_KEY_INJECT = process.env.PROXY_ANONYMOUS_KEY || null;
 
+// ── Inject-value helpers ─────────────────────────────────────────────────
+// Escape a value for safe inclusion in a plist <string>…</string> body.
+function xmlEscape(v) {
+  return String(v).replace(/&/g, "&amp;").replace(/</g, "&lt;").replace(/>/g, "&gt;").replace(/"/g, "&quot;").replace(/'/g, "&apos;");
+}
+// Validate an injected service value: no control chars (a newline would inject a
+// rogue systemd Environment= directive; other control chars corrupt the unit/plist).
+// Spaces are allowed — filesystem paths (CLAUDE_BIN) may legitimately contain them.
+function assertSafeInjectValue(name, v) {
+  if (v == null) return v;
+  if (/[\x00-\x1f]/.test(String(v))) {
+    console.error(`FATAL: ${name} contains a newline or control character — refusing to write it into the service unit.`);
+    process.exit(1);
+  }
+  return v;
+}
+
+// Validate all three INJECT values before they are written into any service unit.
+assertSafeInjectValue("CLAUDE_BIN", CLAUDE_BIN_INJECT);
+assertSafeInjectValue("OCP_ADMIN_KEY", OCP_ADMIN_KEY_INJECT);
+assertSafeInjectValue("PROXY_ANONYMOUS_KEY", PROXY_ANON_KEY_INJECT);
+
 // ── Models: derived from models.json (single source of truth) ──────────
 const modelsConfig = JSON.parse(readFileSync(join(__dirname, "models.json"), "utf-8"));
 
@@ -117,18 +141,26 @@ try {
 }
 
 // Check claude auth (quick test)
-try {
-  const out = execSync('claude -p --output-format text --no-session-persistence -- "ping"', {
-    encoding: "utf-8",
-    timeout: 30000,
-    env: { ...process.env, CLAUDECODE: undefined, ANTHROPIC_API_KEY: undefined, ANTHROPIC_BASE_URL: undefined, ANTHROPIC_AUTH_TOKEN: undefined },
-  }).trim();
-  if (out.length > 0) {
-    log(`Claude CLI authenticated (test response: "${out.slice(0, 40)}...")`);
+// NOTE: This probe uses `claude -p` (sdk-cli spawn). After the 2026-06-15 Anthropic billing
+// split, every `claude -p` call draws from the Agent SDK credit pool rather than the
+// Pro/Max subscription. Re-running setup after 6/15 will consume one metered credit.
+// Set OCP_SKIP_AUTH_TEST=1 to skip this probe (auth is still validated at first real request).
+if (process.env.OCP_SKIP_AUTH_TEST === "1") {
+  warn("OCP_SKIP_AUTH_TEST=1 — skipping claude auth probe (will be validated at first request).");
+} else {
+  try {
+    const out = execSync('claude -p --output-format text --no-session-persistence -- "ping"', {
+      encoding: "utf-8",
+      timeout: 30000,
+      env: { ...process.env, CLAUDECODE: undefined, ANTHROPIC_API_KEY: undefined, ANTHROPIC_BASE_URL: undefined, ANTHROPIC_AUTH_TOKEN: undefined },
+    }).trim();
+    if (out.length > 0) {
+      log(`Claude CLI authenticated (test response: "${out.slice(0, 40)}...")`);
+    }
+  } catch (e) {
+    warn(`Claude CLI auth test failed: ${e.message.slice(0, 100)}`);
+    warn("Make sure you're logged in: claude login");
   }
-} catch (e) {
-  warn(`Claude CLI auth test failed: ${e.message.slice(0, 100)}`);
-  warn("Make sure you're logged in: claude login");
 }
 
 // Check openclaw config (optional — OCP runs standalone without OpenClaw)
@@ -302,7 +334,7 @@ if (OPENCLAW_PRESENT) {
     `║  Aider / OpenClaw) at:                                       ║`,
     `║    http://${BIND_ADDRESS}:${String(PORT)}/v1${" ".repeat(Math.max(0, 47 - BIND_ADDRESS.length - String(PORT).length))}║`,
     `║                                                              ║`,
-    `║  See README § "Client Setup" for per-IDE instructions.       ║`,
+    `║  See docs/lan-mode.md for per-IDE client setup.              ║`,
     `║                                                              ║`,
   );
 }
@@ -358,7 +390,9 @@ if (!DRY_RUN) {
   // and "ocp-proxy" keeps the proxy invisible to that heuristic.
   const OCP_HOME = join(HOME, ".ocp");
   const ocpLogsDir = join(OCP_HOME, "logs");
-  if (!existsSync(ocpLogsDir)) mkdirSync(ocpLogsDir, { recursive: true });
+  // mode 0700: with `recursive`, this call can create ~/.ocp ITSELF on a fresh install, and
+  // without an explicit mode that parent lands at the umask default (world-listable 0755).
+  if (!existsSync(ocpLogsDir)) mkdirSync(ocpLogsDir, { recursive: true, mode: 0o700 });
 
   // Uninstall legacy service names if present (upgrade path)
   if (platform === "darwin") {
@@ -401,17 +435,17 @@ if (!DRY_RUN) {
   <key>EnvironmentVariables</key>
   <dict>
     <key>CLAUDE_PROXY_PORT</key>
-    <string>${PORT}</string>
+    <string>${xmlEscape(PORT)}</string>
     <key>CLAUDE_BIND</key>
-    <string>${BIND_ADDRESS}</string>
+    <string>${xmlEscape(BIND_ADDRESS)}</string>
     <key>CLAUDE_AUTH_MODE</key>
-    <string>${AUTH_MODE_CONFIG}</string>${CLAUDE_BIN_INJECT ? `
+    <string>${xmlEscape(AUTH_MODE_CONFIG)}</string>${CLAUDE_BIN_INJECT ? `
     <key>CLAUDE_BIN</key>
-    <string>${CLAUDE_BIN_INJECT}</string>` : ""}${OCP_ADMIN_KEY_INJECT ? `
+    <string>${xmlEscape(CLAUDE_BIN_INJECT)}</string>` : ""}${OCP_ADMIN_KEY_INJECT ? `
     <key>OCP_ADMIN_KEY</key>
-    <string>${OCP_ADMIN_KEY_INJECT}</string>` : ""}${PROXY_ANON_KEY_INJECT ? `
+    <string>${xmlEscape(OCP_ADMIN_KEY_INJECT)}</string>` : ""}${PROXY_ANON_KEY_INJECT ? `
     <key>PROXY_ANONYMOUS_KEY</key>
-    <string>${PROXY_ANON_KEY_INJECT}</string>` : ""}
+    <string>${xmlEscape(PROXY_ANON_KEY_INJECT)}</string>` : ""}
   </dict>
   <key>RunAtLoad</key>
   <true/>

test-env.mjs

@@ -0,0 +1,26 @@
+// Imported FIRST by test-features.mjs, before keys.mjs, so this runs before anything can open
+// the key store. ESM hoists imports and evaluates them in order, so a `process.env.X = ...`
+// statement in the test's own body would run too late — hence a separate module.
+//
+// Why this exists: `npm test` used to write real, UNREVOKED api_keys rows into the operator's
+// live ~/.ocp/ocp.db (the same database the running server reads) — two per run, unbounded.
+// It also made the suite racy: two concurrent runs (e.g. review worktrees) shared one file, so
+// `listKeys()` could miss "test-user-1" and the `in` check would throw on undefined.
+import { mkdtempSync, rmSync } from "node:fs";
+import { tmpdir } from "node:os";
+import { join } from "node:path";
+
+export const TEST_OCP_DIR = mkdtempSync(join(tmpdir(), "ocp-test-"));
+
+// BOTH are required. keys.mjs honors OCP_DIR_OVERRIDE only when NODE_ENV === "test", so neither
+// var alone redirects anything — a stray OCP_DIR_OVERRIDE in a production env is inert without
+// NODE_ENV=test alongside it. (A daemon OCP launches never carries either: the service units and
+// the `ocp` restart fallback strip both — see plist-merge NEVER_PRESERVE / keys.mjs's comment.)
+process.env.NODE_ENV = "test";
+process.env.OCP_DIR_OVERRIDE = TEST_OCP_DIR;
+
+// Remove the scratch store on exit. Without this the fix would trade unbounded growth in
+// ~/.ocp/ocp.db for unbounded growth in $TMPDIR — better, but still litter.
+process.on("exit", () => {
+  try { rmSync(TEST_OCP_DIR, { recursive: true, force: true }); } catch { /* best effort */ }
+});