docs(issues): record four session findings and close the Dev Drive fleet inventory

[BigSimmo]

Summary

Five immutable inbox requests from the issue #2270 Part A session — four new findings and one update. No canonical ledger edit; docs/outstanding-issues.md is untouched and these reconcile later.

• verify:phone-chrome can exit 0 having run no browser tests at all (P2). Invoked bare on a clean main, its changed-file selector found nothing phone-chrome-affecting, printed a note to that effect, ran two docs checks and exited 0 — indistinguishable from a full pass. Caught while running the A2 gates for Local-only follow-ups after #2253 / #2266 / #2268 — browser gates, orphaned commits, dead reconcile branches, Dev Drive #2270; reporting it as green would have been a false green of exactly the kind AGENTS.md warns about.
• tests/gate-receipts.test.ts is permanently two-red on the Windows workstation (P3). Two cases chmodSync(path, 0o755) and assert the working-tree mode changes the signature. On this ReFS Dev Drive core.fileMode is false and chmod is a no-op, so they cannot pass. Reproduced on unmodified main at 73b1e71a0. This also stops verify:pr-local before it reaches build, check:bundle-budget and the RAG fixture checks.
• The Claude push-format-guard hook blocks every push from a linked worktree (P3). It compares core.hooksPath against the worktree root instead of the common git dir, concludes the repository pre-push hook is unwired when it is in fact wired and running, then fails a whole-tree Prettier check on a git-ignored file that could never be part of the push.
• Seven registered worktree directories are empty while still being the cwd of seven live sessions (P3). Pre-existing, not caused by that session — the inventory counted zero files in each before any removal was attempted, and all seven then refused removal because a live process holds them.

#6GW95D update — the fleet inventory this row asked for is complete, and the cleanup is deliberately deferred. Eight roots scanned: 208 checkouts of this repository, of which 92 are registered worktrees and 116 are separate full clones with their own object databases — and none of those 116 is covered by scripts/clean-worktree.mjs at all, which is the finding that matters for future sweeps. Zero unregistered worktrees, zero stale gitdir pointers. clean:worktree --merged --squashed --dry-run identified 23 landed candidates totalling 9.30 GB. Capacity is no longer the driver: D: is 80 GB with 53 GB free.

The owner halted the removal pass mid-run and all 13 removed worktrees were restored — branch, head and clean tree verified individually, dependencies restored by byte-identical copy. Five live chats had had their working directory deleted underneath them. Two process lessons are recorded on the row: a git refused / EPERM removal means a live process holds that directory and is a stop, never something to retry on a later pass; and the candidate list must be shown and approved before any deletion.

Verification

• npm run verify:pr-local

Documentation-only scope: the five added files are immutable JSON request records under docs/outstanding-issues-inbox/, which the repository treats as append-only and merge-safe. npx prettier --check docs/outstanding-issues-inbox/*.json reports All matched files use Prettier code style!, and scripts/ledger-inbox.mjs validated each request against its schema at creation time — an invalid payload cannot be written.

Verification not run: the executable gates (lint, typecheck, test, build) were not run for this change, because it adds no executable code, no test, and no configuration — there is no plausible failure path for them to detect. The full local suite was run earlier in the same session against the same base and is two-red for the pre-existing Windows chmod reason described above, which is one of the findings filed here.

• UI verification not run: no UI, routing, styling, or browser behaviour is touched.

Risk and rollout

• Risk: None. Five new files, no modifications and no deletions. Ordinary branches never edit the canonical ledger, so this cannot conflict with a concurrent reconcile.
• Rollback: git revert this single commit.
• Provider or production effects: None.
• RAG impact: none.

Notes

Run npm run issues:reconcile from a dedicated fresh-base branch once this lands, to apply the five requests to docs/outstanding-issues.md.

[supabase]

This pull request has been ignored for the connected project sjrfecxgysukkwxsowpy because there are no changes detected in supabase directory. You can change this behaviour in Project Integrations Settings ↗︎.

---

Preview Branches by Supabase.
Learn more about Supabase Branching ↗︎.

[coderabbitai]

Warning

Review limit reached

You’ve reached a temporary PR review limit under our Fair Usage Limits Policy.

Your current included review allowance is based on your included PR review attempts over the past 7 days.

Next review available in:57 minutes

Limit details: You’ve used the included review currently available. Your 90 included PR review attempts over the past 7 days set your current allowance at 1 review per hour.

Your organization has reached its usage spending cap. Adjust your spending cap in the billing tab.

How can I continue?

Wait for the limit to reset, then comment @coderabbitai review or push new commits to the PR.

An organization admin can change what happens after included review limits in Billing.

How do review limits work?

CodeRabbit enforces per-developer PR review limits within each organization.

For paid Pro and Pro+ reviews, CodeRabbit uses a developer's included PR review attempts over the past 7 days to set the current hourly allowance. At typical activity levels, the full plan allowance applies. Higher sustained activity can lower the allowance until earlier attempts leave the 7-day window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 669ffcc1-126d-4034-8638-036b8efb4827

📥 Commits

Reviewing files that changed from the base of the PR and between d617473 and 5b1c116.

📒 Files selected for processing (6)

• data/outstanding-issues-snapshot.json
• docs/outstanding-issues-inbox/2a4e6a53-2fc9-4142-8d8f-e259cb985593.json
• docs/outstanding-issues-inbox/7f369171-54d4-4076-92b8-bf44aab3511d.json
• docs/outstanding-issues-inbox/a5310065-4f20-4036-8992-bc94f7e7e7d2.json
• docs/outstanding-issues-inbox/b7b5fb83-303c-43d4-89dd-228c03bb820e.json
• docs/outstanding-issues-inbox/b7d514e3-a9e4-4178-a210-53ee730de680.json

---

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

CI triage

CI failed on this PR. Automated classification of the 2 failed job(s):

• Static PR checks — needs investigation: inspect the failing step and uploaded diagnostics; rerun only after classifying the cause.
• PR required — needs investigation: inspect the failing step and uploaded diagnostics; rerun only after classifying the cause.

Compared with main CI run #13186 (failure).

Classification is evidence routing, not permission to ignore a failure. Exact quarantined Playwright identities remain governed by the flake ledger.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:13e62ef703

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[BigSimmo]

Final head SHA:636ba2628d9826c58715e5942ac716bc8975748d

Blocker found and disposition

mergeable_state: blocked was not a content conflict — git merge-tree --write-tree origin/main 13e62ef7... reported a clean tree merge. The real blocker was the required static-pr job's "Outstanding-issues ledger integrity" step: this PR adds 5 new pending inbox requests without regenerating data/outstanding-issues-snapshot.json, so npm run check:outstanding-issues failed with counts.pending: committed 32 vs regenerated 37.

Fix: ran node scripts/generate-outstanding-issues-snapshot.mjs (the exact command the check's own failure message names) and committed the regenerated snapshot. No canonical ledger edit — docs/outstanding-issues.md is still untouched, exactly as this PR's description intends.

Sync / merge-tree status

While preparing the fix, the branch's remote head moved (the PR owner merged origin/main into it directly — 1894711d, bringing in an unrelated Codex Cloud Python setup change). Merged that into my fix branch with a plain git merge origin/main (clean, no conflicts) rather than rebasing, then re-verified the snapshot was still in step before pushing.

Targeted verification performed

• npm run check:outstanding-issues → all sub-checks pass, [snapshot] in step with data/outstanding-issues-snapshot.json (78 open, 37 pending)
• npm run docs:check-inventory → pass
• npx prettier --check data/outstanding-issues-snapshot.json docs/codex-cloud.md docs/scripts-index.md → pass

No provider-backed checks run.

Required-CI status (at this exact head)

Run 32575061565 — green so far: Static PR checks (including Ledger write discipline / Run ledger write discipline / Outstanding-issues ledger integrity, all passing now), Build, Production UI critical, Lighthouse budget, Safety and config checks. Still legitimately in progress: Unit coverage, Production UI (1/2/3). PR required has not reported yet because it waits on those.

Review-thread status

3 unresolved threads found, all from the automated Codex reviewer:

• Resolved — the P1 "Regenerate the outstanding-issues snapshot" thread, replied with the fixing commit and resolved, since it's exactly the fix above.
• Left open (2 P2 threads) — both point out real content errors inside two of the five new immutable inbox request JSON files (a git worktree add command that's missing the required -f override, and an internal arithmetic/classification inconsistency in the fleet-inventory update record). I verified both findings are correct, drafted fixes, then backed them out: editing recorded finding content in files this PR's own description treats as append-only/immutable is a judgment call about this repo's inbox convention, not a mechanical CI-unblocking fix, so I left both open with a reply explaining exactly what's wrong and what the fix would be, for the PR owner to apply or authorize.

Residual risk / required human action

1. Decide whether to correct the two flagged inbox records before or after this PR lands (see the two open threads).
2. Let Unit coverage and the three Production UI shards finish; PR required will report once they do.
3. A ledger record for this fix (npm run ledger:append) is written locally but not yet pushed — deferred to avoid cancelling this in-flight CI run per this repo's own anti-churn policy; it'll be pushed once these checks settle.

Merge left to the user.

UNBLOCKED — CI PENDING

---

Generated by Claude Code

[BigSimmo]

Closing this in favour of a clean re-do. The content is wanted; this branch is not salvageable at reasonable cost.

Why. Three agents pushed to this branch in parallel — the original session, a Codex autofix pass, and a second Claude session — and the result is wrong in ways that matter for a permanent ledger:

• The autofix "corrected" the empty-directory repair command to git worktree add -f <path> <branch> and justified it as "the forced add restores a path that remains registered to that branch". That is false. Those seven directories are not registered — none appears in git worktree list and none has a .git entry. Verified by execution rather than assumption: git worktree add <existing-empty-path> <branch> into an unregistered, empty path with the branch checked out nowhere succeeds, exit 0, 3812 files checked out. The -f is unnecessary and disables the branch-already-checked-out guard.
• The same records still describe those directories as "registered worktree directories", which is the misclassification the review flagged in the first place.

Publishing either statement into docs/outstanding-issues.md would prescribe a repair based on a false premise.

Two structural problems this exposed, which will recur.

1. data/outstanding-issues-snapshot.json is generated, must be regenerated by every PR that queues an inbox request, and regenerates differently depending on what else has landed. So it conflicts whenever any other ledger PR merges first — twice within an hour here, and four of the recent merges to main touched it. Any PR of this shape that does not land quickly will hit the same wall. Filed as its own finding in the replacement.
2. Merging main into the branch pulled unrelated Codex Cloud files (scripts/setup-codex-cloud.sh, scripts/select-codex-cloud-python.sh, tests/codex-cloud-setup.test.ts) into the PR's own diff, which is what put PR policy into Clinical-risk paths require the ## Clinical Governance Preflight section on a documentation-only change.

Replacement: same five findings plus the snapshot-collision one, re-derived from origin/main, in one small commit, with the corrected and execution-verified text. Nothing is lost — none of these records ever reached docs/outstanding-issues.md.

---

Generated by Claude Code