[ja573]

Integrates feature/worker @ 0b5ceeee (ADR-0012, Amendment 14, landed by #960) into develop. This is THOTH-ASYNC-01-INTEG-01, granted on #958 (comment 6765414073).

What it does

• Merge commit abde91d4 merged develop fc8f90a8 into feature/worker 0b5ceeee.
  • The only conflict was CHANGELOG.md. It was resolved to develop's file plus one [Unreleased] line for ADR-0012, following the 2026-11-07 changelog style instead of the D5 union. This clears OPUS-03, the mis-indented line.
• First refresh (2026-11-10): develop moved to cab25eba when MET-WP4-WIDGET-ADR: record approved ADR-0014 public widget metrics #1009 (ADR-0014) landed. Merge commit ffa7301a merged it into this branch.
  • The one conflict was the decision register's Last updated: line. It was resolved by hand to develop's 2026-11-10.
  • Every other path is the automatic merge, and CHANGELOG.md merged cleanly.
  • refresh_check rates the refresh NOT PURE because of that line. Both CRITICAL reviewers approved the refresh delta.
• Second refresh (2026-11-10): the CTO held this PR until Metrics B2 (MET-WP4-03C-B2: all-time metricDashboard history (no calendar cap, YEAR grain, optional startDate) #1010) landed. It landed as af6719f6, and merge commit 9e584d6c merged it into this branch.
  • The one conflict was CHANGELOG.md. Both sides' [Unreleased] entries are kept (D5).
  • refresh_check rates the refresh PURE. Both CRITICAL reviewers approved the refresh delta.
• The diff against develop is documentation only:
  • ADR-0012;
  • the Superseded in part by headers of ADR-0008 and ADR-0010;
  • the decision-register row;
  • the A14 implementation report;
  • the CHANGELOG line.
• It changes no source, schema, migration or workflow file.

Not in scope

• The CLOSEOUT-01 reconciliation (OPUS-04, OPUS-06, LOW-01, LOW-02, N-01, N-02, register row). It follows on develop under its own spec.

Gates

• Head 9e584d6c. Fresh Opus 5.5 max and Fable reviews of the second refresh delta are APPROVED. CI is green at the head and at develop af6719f6, and land_check returns PASS.
• AWAITING a CTO MERGE AUTH bound to 9e584d6c. It must ratify the hand-resolved refresh ffa7301a and classify the ADR-0008/0010 header lines. After that, a merge-commit landing with the expected head (merge tree d6c8193a while develop is af6719f6).

---

Generated by Claude Code

[ja573]

REVIEW THOTH-ASYNC-01-INTEG-01 @ abde91d: APPROVED

• Reviewer: Claude Opus 5.5, effort max; fresh, independent, read-only (CRITICAL Opus leg). Refs re-verified 11:09Z: develop fc8f90a8, feature/worker 0b5ceeee.
• Checked: parents 0b5ceeee + fc8f90a8, Javier Arias, no AI trailers on any of the 54 incoming commits. Head tree equals git merge-tree of develop and 0b5ceeee on every path except CHANGELOG.md (the only conflict); worker documents byte-identical to 0b5ceeee (ADR-0012 blob 2d3f62f6); all 181 develop-changed paths keep develop's version. CHANGELOG.md = develop's file plus ### Added and one house-format line. land_check.py (develop's copy): PASS. Docs only. CI green.
• D5: parallel-landing applies the union only to refresh merges, and the GRANT rules it out here; a union would have injected 23 bullets into the released 1.9.0 section and left [Unreleased] empty.
• F-1 (P2): on landing ADR-0012 becomes repository-authoritative by itself, while the register row, ADR-0012 §12 and its header still say otherwise. Accepted as stale-by-design: LANDED must state authority from the merge, and CLOSEOUT-01 follows promptly.
• F-2 (P3): after landing, ADR-0012 changes use the factual-clarification path or a superseding ADR. F-3 (P3): citation could be [1006]; settle at release. F-4 (P3): MERGE AUTH should name Ready-for-review, exact SHA abde91d4, a true merge commit (tree 8c494dab) and prompt landing. F-5 (P3): ledger note on early ADR-only integration and the future of feature/worker.

---

Generated by Claude Code

[ja573]

REVIEW THOTH-ASYNC-01-INTEG-01 @ abde91d: APPROVED

• Reviewer: Claude Fable (session model ID claude-fable-5-2), session-default effort; fresh, independent, read-only (CRITICAL Fable leg).
• Checked: PR draft at abde91d4, parents 0b5ceeee then fc8f90a8, Javier Arias, no trailers; neither tip moved. Head tree equals the automatic merge except CHANGELOG.md, the only conflict. The five worker-side documents are byte-identical to 0b5ceeee; develop did not touch docs/engineering/decisions/ since merge base b44303c2; nothing from develop dropped. CHANGELOG.md = develop's file plus ### Added and one line under [Unreleased]. land_check.py from develop: PASS (with and without --authorized). Six paths, all docs. Four CI runs at the head, all green.
• F-1 (P2): landing makes the register row, ADR-0012 §12 and header line 7 false until CLOSEOUT-01. Not blocking: say so in MERGE AUTH and LANDED and run CLOSEOUT-01 promptly; afterwards any byte change to ADR-0012 is a CTO-classified factual clarification naming blob 2d3f62f6.
• F-2 (P3): the line could cite 1006 or [960], [1006]; amend at the next CHANGELOG touch. F-3 (P3): refresh_check.py rates the head NOT PURE, as expected; MERGE AUTH binds to abde91d4 directly and names the non-union resolution.
• INFO: any head change fails the land-check; feature/worker will conflict again unless fast-forwarded after landing (own grant); delete the integration branch after landing; 12 broken relative links in the implementation report pre-date this merge.

---

Generated by Claude Code

[chatgpt-codex-connector]

Codex Review Summary: ✅ Completed

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-11-10T23:05:43.138115Z	9e584d6	New commits

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" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: abde91d4a6

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".

[ja573]

REVIEW THOTH-ASYNC-01-INTEG-01 @ ffa7301: APPROVED (refresh delta of abde91d)

• Reviewer: Claude Opus 5.5, effort max. A fresh, independent, read-only CRITICAL review of the refresh delta under parallel-landing.md section 6. This is the Opus leg; it did not see the other leg's work.
• Faithfulness:
  • Tree d54c59e9 equals git merge-tree --write-tree abde91d4 cab25eba at every path except the decision register.
  • In the register, the only change made by hand is the conflicting Last updated: line, resolved to develop's 2026-11-10. git show --remerge-diff shows only that hunk, and nothing from either side was dropped.
  • The PR's own documents are byte-identical to abde91d, including ADR-0012 blob 2d3f62f6.
• Resolution: 2026-11-10 is correct. The register records the date of its latest content edit, and the CTO-approved #1008 spec prescribed this value. Keeping 2026-10-30 would back-date the register.
• CHANGELOG: merged cleanly. [Unreleased] has ### Added (#960) and then ### Changed (#1008), in house format, with no duplicates.
• Scripts: land_check returns PASS. refresh_check returns NOT PURE, for the register line only.
• Scope: the same six documentation paths as before, and docs_only=true.
• CI: all 4 runs at ffa7301 are green, and develop cab25eb is green.
• F-1 (P2): the PR's Superseded in part by header lines in ADR-0008 and ADR-0010 have no CTO factual-clarification record. #1008 GRANT line (b) shows the same kind of edit to ADR-0001 being classified. Fix: the MERGE AUTH classifies both. The head does not move.
• F-2 (P3): the ADR-0008 and ADR-0010 register rows do not yet note the partial supersession (OPUS-04). This stays with CLOSEOUT-01.
• F-3 (P3): the #958 state block and the PR body still describe abde91d. Update both.
• INFO:
  • The refresh followed a land_check FAIL. The MERGE AUTH should ratify it by name and bind to ffa7301.
  • The expected merge tree is d54c59e9 while develop stays at cab25eb.
• Report: project file async-adr-01/INTEG-01-refresh-review-opus-max-ffa7301a.md.

---

Generated by Claude Code

[ja573]

REVIEW THOTH-ASYNC-01-INTEG-01 @ ffa7301: APPROVED (refresh delta of abde91d)

• Reviewer: Fable (claude-fable-5-2), effort high. Fresh, independent and read-only; this is the Fable leg of the CRITICAL pair. It did not see the other leg's work.
• Identity: the head is ffa7301, with parents abde91d and then cab25eb. Author and committer are Javier Arias, with no trailers. develop is still at cab25eb.
• Faithfulness:
  • The head tree differs from git merge-tree --write-tree abde91d4 cab25eba at one path only, the register (566 entries compared).
  • The register is byte-identical to git merge-file -p --theirs, so the conflict hunk is the only choice made by hand.
  • Both new rows are intact, and so is develop's change to the ADR-0001 row.
• Resolution: 2026-11-10 is correct. All 15 earlier changes to the line carry the date of the newest register edit.
• CHANGELOG: the result equals the automatic merge: one ### Added line (#960), then one ### Changed line (#1008).
• Coherence: nothing #1009 changed interacts with ADR-0012 or its report. ADR-0001's new supersession header follows the pattern of this PR's ADR-0008 and ADR-0010 headers. The CLOSEOUT-01 items are byte-unchanged.
• Scripts: land_check returns PASS. refresh_check returns NOT PURE, for the register line only.
• Scope: the same six documentation paths, +19088 −2.
• CI: 4 of 4 runs green at ffa7301.
• F-1 (P2): the refresh was made after a land_check FAIL (a conflict outside CHANGELOG.md). The quickstart says to post HOLD in that case, and the GRANT names one merge. Fix: the MERGE AUTH names ffa7301 and ratifies the refresh made by hand.
• F-2 (P3): the #958 state block and the PR body still describe abde91d.
• F-3 (P3): parallel-landing.md uses "reviewed in full" ambiguously. Both readings are satisfied here.
• F-4 (P3): the old Codex P1 thread has been answered but is still unresolved.
• Report: project file async-adr-01/INTEG-01-refresh-review-fable-ffa7301a.md.

---

Generated by Claude Code

[ja573]

REVIEW THOTH-ASYNC-01-INTEG-01 @ 9e584d6: APPROVED (refresh delta of ffa7301)

• Reviewer: Claude Opus 5.5, effort max. A fresh, independent, read-only CRITICAL review of the second refresh delta under parallel-landing.md section 6. This is the Opus leg; it did not see the other leg's work.
• Identity: the head is 9e584d6, with parents ffa7301 and then af6719f. Author and committer are Javier Arias, with no trailers on any of the PR's 56 commits. develop is still at af6719f, the #1010 landing merge.
• Faithfulness:
  • Every path except CHANGELOG.md equals git merge-tree --write-tree ffa7301a af6719f6.
  • A git merge-file --union rebuild of CHANGELOG.md matches the head blob byte for byte. --remerge-diff shows only the three conflict markers removed, so nothing from either side was dropped.
  • The PR's five documents are unchanged from ffa7301, including ADR-0012 blob 2d3f62f6.
• CHANGELOG: the PR's diff against develop is exactly the one #960 line, in house format, under ### Added. Every #1010 and #1008 line appears once.
• Coherence: nothing in the PR's documents relies on anything #1010 changed, and the reverse also holds.
• Scripts: land_check returns PASS. refresh_check returns PURE.
• Scope: the same six documentation paths (+19087 −2), with no source, schema, migration, workflow or config change.
• CI: all 4 runs at the head are green, and develop's CI at af6719f is green.
• F-1 (P3): the PR body and the #958 state block lagged this refresh when read. Update both.
• F-2 (INFO): develop's [Unreleased] has two ### Changed headings. They came from the #1010 landing, not from this PR, which does not make them worse. Tidy them on develop separately.
• F-3 (INFO): refresh_check accepts the union in either side order. This head uses the correct order, which puts the #960 line under ### Added. Consider having the landing rules check where each merged line lands.
• F-4 (INFO): the MERGE AUTH must bind to 9e584d6. While develop stays at af6719f, the landing tree is d6c8193a. The two P2 points from the ffa7301 reviews still go to the MERGE AUTH, and this refresh left their bytes unchanged. If anything lands first (for example #1011), expect another CHANGELOG-only refresh and two more delta reviews.
• Report: project file async-adr-01/INTEG-01-refresh2-review-opus-max-9e584d6c.md.

---

Generated by Claude Code

[ja573]

REVIEW THOTH-ASYNC-01-INTEG-01 @ 9e584d6: APPROVED (refresh delta of ffa7301)

• Reviewer: Fable (claude-fable-5-2). Fresh, independent and read-only; this is the Fable leg of the CRITICAL pair. It did not see the other leg's work.
• Identity: the head is 9e584d6, with parents ffa7301 and then af6719f. Author and committer are Javier Arias, with no trailers. af6719f is still the tip of develop.
• Faithfulness:
  • The head tree differs from git merge-tree --write-tree ffa7301a af6719f6 at CHANGELOG.md only.
  • A git merge-file --union rebuild is byte-identical to the head blob fd2e8bd1. --remerge-diff shows only the three conflict markers removed.
  • The other five PR paths carry ffa7301's blobs exactly, including ADR-0012 at 2d3f62f6.
• CHANGELOG: the PR's diff against develop is the single #960 line, in house format, under ### Added. Every #1010 and #1008 line is present once.
• Coherence: the five PR documents have no hit for metricDashboard, 366, METRIC_READ_BUSY, contract-register or statement_timeout. Every deadline hit is about job deadlines. #1010's diff mentions none of the ADRs or the job engine.
• Scripts: land_check returns PASS and refresh_check returns PURE. PURE rests on D5: with --no-changelog-union it would be NOT PURE.
• Scope: the same six documentation paths, +19087 −2. None of #1010's code paths is in the diff.
• CI: 4 of 4 runs green at the head. Both push runs at af6719f are green.
• F-1 (INFO): the refresh was pushed while develop's post-merge CI at af6719f was still running. It is green now. Cite both push runs in the MERGE AUTH basis and the LANDED evidence.
• F-2 (P3): the PR body and the #958 state block lagged this refresh when read. Update both.
• F-3 (INFO): the two ### Changed headings under [Unreleased] come from develop's clean automatic merge of #1010. This PR neither created them nor makes them worse. Tidy them at the next CHANGELOG touch or the release cut.
• F-4 (INFO): the MERGE AUTH must name 9e584d6. The landing tree must be d6c8193a while develop stays at af6719f. The two P2 points from the ffa7301 reviews still go to the MERGE AUTH.
• F-5 (INFO): #1011 adds a line to the same ### Added block. Whichever lands second should expect a CHANGELOG-only refresh.
• F-6 (INFO): the refresh commit is unsigned, like abde91d and ffa7301.
• Report: project file async-adr-01/INTEG-01-refresh2-review-fable-9e584d6c.md.

---

Generated by Claude Code