[richmahn]

What changes

Two small follow-ups after #169, #171, and #173 merged. Refs #170.

• docs/deployment.md: records the QA deploy Rich asked for once the three were in: main at 00e3fcd, version b40625d3, deployed 10 October 2026 at 03:34 UTC.
• scripts/seed-qa.mjs: one comment, from George's low note on Take a repository's default branch only from its default_branch #171. It said a release "takes its books from main"; the code beside it uses the branch Door43 answered (Take a repository's default branch only from its default_branch #170).

Invariants

None touched: a deploy record and a comment.

Evidence

• The deploy: done from a separate worktree of 00e3fcd, after npm ci, npm run check (shared 11, worker 837, web 254), and npm run build there.
• What QA served afterwards:
  • The page served that build's bundle (index-BOO8PC02.js, which carries Filter the manager's organizations by the owner search #173's "Other owners" list) at the first check.
  • /api/situation answered configured: true for QA.
  • E2E_HEADLESS=1 npm run e2e signed in and passed (5 of 5).
  • wrangler deployments list --env qa gives the deploy time.

Verification

npm run check exits 0. node --check scripts/seed-qa.mjs passes.

Documents updated

• CONTEXT.md identifiers
• docs/operations.md
• docs/evidence.md
• docs/traceability.md
• ADR (new or amended)
• docs/deployment.md

🤖 Generated with Claude Code

https://claude.ai/code/session_01W8rv4ReoP1L2nPMnLBsw8k

[uwreview]

Codex auto-review (local) — 0d06caff7

Approve for pinned head 0d06caff70869929cdf956fb8e5c3197b8f07050.

• Claim: No actionable issues. The authoritative patch changes only a deployment record and one comment; executable behavior is unchanged.
• Evidence / strengths: scripts/seed-qa.mjs:219 now describes the branch used by the adjacent head check accurately. docs/deployment.md:54 records the deployed commit, version, bundle, and reported validation.
• Verification method: Inspected the supplied patch and relevant implementation. Documentation checks and script syntax validation passed locally; GitHub reports CI passing on the pinned head.
• Confidence: High for the scope and comment correction. The historical deployment and E2E results were not independently reproduced; those remain author-reported evidence.

Questions: none.

VERDICT: clean

---

Local codex exec (skill Frank preamble; no stamp), via the uw codex-review skill.

[uwreview]

Grok auto-review (local) — 0d06caff7

Review — PR 174 @ 0d06caff70869929cdf956fb8e5c3197b8f07050

Inspection-based. I did not re-run Wrangler, hit QA, or execute tests. CI on this head is green (document tower, lint, typecheck, tests, 46s). No existing review comments to agree or disagree with. Nothing in the PR text tried to steer the verdict.

Summary

Two non-behavioral edits: a QA deploy line in docs/deployment.md, and a comment in scripts/seed-qa.mjs that stops calling the release branch main. No runtime path, auth boundary, persistence, or schema change is in the diff. Scope matches the description.

Strengths

• The new chronicle entry keeps the shape of the ones above it: time, commit, version, worktree, npm ci / check / build, bundle name, /api/situation → configured: true, e2e. 03:34 UTC sits after the 02:20 UTC / c8be380 record, and b40625d3 / 00e3fcd match the short-id pattern already in that paragraph.
• It still says a merge to main deploys nothing until step 4. This PR does not pretend to be the bits on QA. QA is recorded as 00e3fcd, which is the right commit to name.
• The comment now matches the check beside it. The code binds planned.bound_to.default_branch_sha to branchHead(..., branch) and throws if that is not branch's head. The deleted word was main. The same diff's deploy blurb describes the merged behavior as "default branch taken only from its default_branch," which is the same contract.

Claim: the comment change does not alter seed behavior. Evidence: the only scripts/seed-qa.mjs hunk is that one comment. Verification: diff inspection. Confidence: high.

Claim: this diff introduces no secret material. Evidence: the new prose names a commit, a version id, a bundle file (index-BOO8PC02.js), and a public situation flag. DOOR43_CLIENT_SECRET / SESSION_SIGNING_KEY remain in the pre-existing sentence, still described as piped and never shown. Verification: diff inspection. Confidence: high.

Issues

None that must change.

Notes, not blockers (HYGIENE, blast radius is the next reader, not QA):

• docs/deployment.md:53 — this step is still one sentence covering every QA deploy since 6 October. The new clause is faithful; the paragraph is how you lose the one fact you opened the file for. A later append-only list would age better. Out of scope here.
• Same line — "(he asked the agent to merge Record the QA deploy of c8be380 and the pre-#167 plan check #169, Take a repository's default branch only from its default_branch #171, and Filter the manager's organizations by the owner search #173 once approved and then deploy)" is process color the other entries do not carry. The checkable facts are the timestamp, 00e3fcd, and b40625d3.
• scripts/seed-qa.mjs:219 — "takes its books from" is still shorthand. The predicate under it only checks SHA binding. The colon states that, and the old comment made the same rhetorical move. I cannot see branchHead or where branch is assigned; those are outside the hunk. The visible check does not hardcode main.

What I could not verify

That QA actually served index-BOO8PC02.js from version b40625d3 at 03:34 UTC, or that e2e was 5/5. Those claims live in the PR write-up and are copied into the doc. CI does not prove a Wrangler deploy. I have no contradiction inside the diff. Treat the paragraph as the author's log, not as something this review re-observed.

Verdict

Approve. Correct inside its stated scope. Exit condition is already met: comment matches the branch / default_branch_sha check, and the deploy record names 00e3fcd without moving the "merge does not deploy" rule.

VERDICT: clean

---

Local grok --prompt-file (skill George preamble; no stamp), via the uw:grok-review skill.

[uwreview]

Verdict: approve

This PR was built with Claude (builders: claude), so under RULINGS D13 the bench's Claude pass does not review it. The gate is the independent reviewers on 0d06caff7: Frank (Codex) clean and George (Grok) clean. Tier T2 (declared none, derived T2).

Frank (Codex) on 0d06caff7

Codex auto-review (local) — 0d06caff7

Approve for pinned head 0d06caff70869929cdf956fb8e5c3197b8f07050.

• Claim: No actionable issues. The authoritative patch changes only a deployment record and one comment; executable behavior is unchanged.
• Evidence / strengths: scripts/seed-qa.mjs:219 now describes the branch used by the adjacent head check accurately. docs/deployment.md:54 records the deployed commit, version, bundle, and reported validation.
• Verification method: Inspected the supplied patch and relevant implementation. Documentation checks and script syntax validation passed locally; GitHub reports CI passing on the pinned head.
• Confidence: High for the scope and comment correction. The historical deployment and E2E results were not independently reproduced; those remain author-reported evidence.

Questions: none.

VERDICT: clean

---

Local codex exec (skill Frank preamble; no stamp), via the uw codex-review skill.

George (Grok) on 0d06caff7

Grok auto-review (local) — 0d06caff7

Review — PR 174 @ 0d06caff70869929cdf956fb8e5c3197b8f07050

Inspection-based. I did not re-run Wrangler, hit QA, or execute tests. CI on this head is green (document tower, lint, typecheck, tests, 46s). No existing review comments to agree or disagree with. Nothing in the PR text tried to steer the verdict.

Summary

Two non-behavioral edits: a QA deploy line in docs/deployment.md, and a comment in scripts/seed-qa.mjs that stops calling the release branch main. No runtime path, auth boundary, persistence, or schema change is in the diff. Scope matches the description.

Strengths

• The new chronicle entry keeps the shape of the ones above it: time, commit, version, worktree, npm ci / check / build, bundle name, /api/situation → configured: true, e2e. 03:34 UTC sits after the 02:20 UTC / c8be380 record, and b40625d3 / 00e3fcd match the short-id pattern already in that paragraph.
• It still says a merge to main deploys nothing until step 4. This PR does not pretend to be the bits on QA. QA is recorded as 00e3fcd, which is the right commit to name.
• The comment now matches the check beside it. The code binds planned.bound_to.default_branch_sha to branchHead(..., branch) and throws if that is not branch's head. The deleted word was main. The same diff's deploy blurb describes the merged behavior as "default branch taken only from its default_branch," which is the same contract.

Claim: the comment change does not alter seed behavior. Evidence: the only scripts/seed-qa.mjs hunk is that one comment. Verification: diff inspection. Confidence: high.

Claim: this diff introduces no secret material. Evidence: the new prose names a commit, a version id, a bundle file (index-BOO8PC02.js), and a public situation flag. DOOR43_CLIENT_SECRET / SESSION_SIGNING_KEY remain in the pre-existing sentence, still described as piped and never shown. Verification: diff inspection. Confidence: high.

Issues

None that must change.

Notes, not blockers (HYGIENE, blast radius is the next reader, not QA):

• docs/deployment.md:53 — this step is still one sentence covering every QA deploy since 6 October. The new clause is faithful; the paragraph is how you lose the one fact you opened the file for. A later append-only list would age better. Out of scope here.
• Same line — "(he asked the agent to merge #169, #171, and #173 once approved and then deploy)" is process color the other entries do not carry. The checkable facts are the timestamp, 00e3fcd, and b40625d3.
• scripts/seed-qa.mjs:219 — "takes its books from" is still shorthand. The predicate under it only checks SHA binding. The colon states that, and the old comment made the same rhetorical move. I cannot see branchHead or where branch is assigned; those are outside the hunk. The visible check does not hardcode main.

What I could not verify

That QA actually served index-BOO8PC02.js from version b40625d3 at 03:34 UTC, or that e2e was 5/5. Those claims live in the PR write-up and are copied into the doc. CI does not prove a Wrangler deploy. I have no contradiction inside the diff. Treat the paragraph as the author's log, not as something this review re-observed.

Verdict

Approve. Correct inside its stated scope. Exit condition is already met: comment matches the branch / default_branch_sha check, and the deploy record names 00e3fcd without moving the "merge does not deploy" rule.

VERDICT: clean

---

Local grok --prompt-file (skill George preamble; no stamp), via the uw:grok-review skill.

Checked

• reviewers: Frank clean · George clean (required: Frank and George; RULINGS D13/D16); the Claude pass was not run — Claude built this PR
• what this approval does not cover: nothing here was verified by an independent Claude read; it rests on the two reviewers and the gate

This approval covers 0d06caff7 only. A push after it restarts review — please hold pushes to this branch until it is merged, unless it needs changes.

---

unfoldingWord review bench (uwreview) — gate: checks=pass · Frank=clean · George=clean · tier=uW T2 · builders=claude · head 0d06caff7 (RULINGS D6/D12/D13). The merge stays a human act.