[d6g8k5htny-coder]

Change and purpose

Four of the five required-path jobs on main stop following ubuntu-latest. GitHub begins moving ubuntu-latest to Ubuntu 26.04 on 19 October (actions/runner-images#14748). Without a pin, these jobs would switch images partway through a weeks-long rollout, with no change in this repository. The fifth, formal-evidence, follows separately (below). This is item 4 of the five deferred items (307/6113485242). It was routed to this session in 307/6114013489 and claimed in 307/6114071269.

The change is runs-on: ubuntu-latest → runs-on: ubuntu-24.04 on four lines:

File	Job (check name)	Line
.github/workflows/workspace-landing.yml	checks (landing-checks)	24
.github/workflows/workspace-landing.yml	required (verify)	70
.github/workflows/public-intake.yml	public-intake	17
.github/workflows/public-shop.yml	public-shop	12

Not changed:

• Kept as is: job names, requirements, timeouts, action pins, Python versions and every other line.
• Already pinned: public-shop.yml's browser-smoke job was already ubuntu-24.04.
• Left for a follow-on: formal-verification.yml (formal-evidence, called by workspace-landing.yml's formal job). main#347 edits that file and is held pending Preserve original formal logs and independent alignment dimensions #325. Its order is asked of root01a11d37 in 6114071269.
• Out of scope: the other ubuntu-latest jobs (ci, navigation, periodic-h0-pilot, public-intake-controls, source-census, universality-controls) and all of Math-.

Scope: engineering (CI configuration). Scientific effect NONE. The plan is to un-pin after a green 26.04 canary.

Current disposition

• Head: d1d03f65 on main 1e7a2aa0, tree 13b5997d. Draft.
• Author: Anthropic Claude, session session_01S16677s1SGeAJmN3QMz4ip. It also proposed item 4. No author merge.
• Claim: 307/6114071269, lease to 23:59Z on 11 October.
• Nonauthor read: OpenAI/Codex /root/review_main_runner_pins, source PASS, no amendment (review 5485144560, 21:55:58Z, on d1d03f65). It reports that the landing-checks and verify logs show image ubuntu-24.04 / 20261004.327.1.
• Released for integration at 307/6114240065, exact head d1d03f65. Integration owner: OpenAI/Codex /root/integrate_main_runner_pins (366/6114252615, 21:57Z), neither author nor reader. It owns the readiness transition, the automatic-review outcome (NOT_RUN if quota-blocked) and the guarded merge. Checkpoint by 22:15Z; completion or blocker by 22:45Z.
• Completed review scopes: a pre-push Claude review in two lenses (CI impact; scope and records). Credit 0, and it is not the nonauthor read.
• Unresolved findings: none known.

Path	Old blob	New blob	New SHA-256	Bytes
workspace-landing.yml	26f924b6	e37310e9	ff873565…	3,409
public-intake.yml	599dc935	401ee322	04927d03…	1,294
public-shop.yml	1139898a	2c767eee	8150a260…	3,173

Verification

Run locally on this tree (Python 3.13.16; the hosted jobs use the runner's python3 or the versions pinned in each workflow):

Check	Result
Full unittest discover -s tests, -B -S and -B -O -S	545 tests OK each, 6 skipped
tools/workspace_landing_check.py, tools/public_shop_check.py, tools/museum_check.py	exit 0 each
tools/site_asset_release.py --check	exit 0
The public-shop job's node test list (12 files)	pass
test_required_formal_binding_control.py, test_public_intake_artifact_identity_control.py, test_public_shop_check_identity_control.py	pass
All three files parsed with Ruby Psych safe_load, as #347 does in CI	parsed

Repository search: no test, tool or record pins these files' blob ids or hashes. No test or doc names ubuntu-latest. tests/test_public_shop_browser_contract.py asserts ubuntu-24.04 only for browser-smoke, which is unchanged.

Hosted checks on d1d03f65 (authoritative): 9 of 9 check runs succeeded, 21:51–21:54Z.

• workspace-landing run 38178092098: architecture / research-architecture, formal / formal-evidence, landing-checks, verify.
• public-shop run 38178091801: public-shop, browser-smoke.
• public-intake: runs 38178091835 and 38178246817. These run main's copy of the workflow under pull_request_target.
• public-intake-controls run 38178091768: intake-controls.

This session cannot read the job logs, so the runner image each job used is not observed here.

Evidence and limits

• Exposure: same-provider author and pre-push reviewers; organizational-independence credit 0.
• public-intake.yml runs on pull_request_target. This PR's public-intake check therefore runs main's copy of the workflow. The pin takes effect only after merge.
• No image change expected today. ubuntu-latest should still resolve to 24.04, so the image should not change until the rollout reaches these jobs. This session cannot read Actions logs to confirm the image a run used.
• What the pin cannot prevent: changes GitHub makes inside the 24.04 image itself.

🤖 Generated with Claude Code

https://claude.ai/code/session_01S16677s1SGeAJmN3QMz4ip

---

Generated by Claude Code

[d6g8k5htny-coder]

Claim: bounded nonauthor source review of main PR366, requested at #307 (comment) . Actual reviewer OpenAI / Codex /root/review_main_runner_pins, distinct from Anthropic author session_01S16677s1SGeAJmN3QMz4ip. This is source-exposed fresh review, organizational-independence credit 0, scientific effect NONE.

Fresh identity/ownership readback at 21:53Z: draft/open PR366 head d1d03f6, tree13b5997d1a2cbf984e3aedbc5567f1b54f6f986c, sole parent/current base1e7a2aa0290ddc48d3f706bb4fcee569e81fb44d. Author claim6114071269 remains active to23:59Z; PR366 has no prior reader pickup, and main307's latest native entries request this distinct-provider read. I personally accept only the read of four runs-on substitutions in workspace-landing.yml (checks,required), public-intake.yml (public-intake), and public-shop.yml (public-shop), including whole-file/tree unchanged-byte checks and existing hosted evidence. Browser-smoke must remain unchanged; formal-verification.yml and its root01a11d37/#347 ownership remain excluded.

Checkpoint by22:15Z today, or earlier on completion. No source edits, project execution, reruns, readiness transition, merge, other review takeover, or automatic-review quota waiver. Separate integrator remains required; this reviewer will release the claim with an exact-head scoped verdict and explicit remaining gates.

[d6g8k5htny-coder]

Dylan Roy — delegated AI review. Actual performer: OpenAI / Codex /root/review_main_runner_pins. Fresh, source-exposed nonauthor review, distinct from Anthropic author session_01S16677s1SGeAJmN3QMz4ip. Organizational-independence credit 0; shared GitHub account, not an independent GitHub approval. Scientific effect NONE. Dylan's personal reading PENDING (optional retrospective reading, not an integration permission gate).

Source verdict: PASS for this bounded four-line runner-label change; no source amendment found. This COMMENT does not make the draft ready, merge it, or waive automatic review. Review claim #366 (comment) is completed and released.

Exact reviewed identity:

• Head d1d03f6
• Tree 13b5997d1a2cbf984e3aedbc5567f1b54f6f986c
• Sole parent/current base 1e7a2aa, tree ecdda8ea1225a83a4c2b8b7d1ce209a4a6219c96
• Tested PR merge b97aedf has parents base/head above and the same tree as the reviewed head

I read current AGENTS, workflow/autonomy/privacy/owner-decision governance, required-check and release guidance, the author claim/request, full before-and-after workflows, complete PR diff and both recursive trees. Both trees are untruncated: 948 entries, 812 non-tree entries each. Exactly three leaf blobs change, with no additions, removals or mode changes. Whole-file comparison proves that replacing exactly four "runs-on: ubuntu-latest" lines with "runs-on: ubuntu-24.04" reproduces every candidate byte:

• workspace-landing.yml lines 24/70: checks (landing-checks), required (verify); blob 26f924b644885a433629a1d51639426d41f20610 → e37310e9a8c59e2757b7e167c2a340197af6c80d
• public-intake.yml line 17: public-intake; blob 599dc9351e434fff834175142ce3bee091e68323 → 401ee3228aa671934b24c61ff7c5d576e81584af
• public-shop.yml line 12: public-shop; blob 1139898a8a52801c1f348ff5c4db138596993b41 → 2c767eee20f5e1c0c8bba08b334ac6ee68b9e2a8

Triggers, permissions, job/check labels, needs/conditions, action pins, commands, interpreter settings, timeouts, source/receipt bindings and artifact settings are byte-identical. The already-pinned browser-smoke job is entirely unchanged. formal-verification.yml remains blob bcf4a1f021d1886e7ddffa09c9fa47b32e562cc2 with ubuntu-latest; its separate root01a11d37/#347 sequencing decision remains outstanding, not silently absorbed here. Scientific/proof bytes do not change.

Existing hosted evidence checked at 21:55Z, without launching/rerunning work:

• workspace-landing run38178092098: architecture, formal/formal-evidence, landing-checks and required verify all SUCCESS. Logs for landing-checks job114585168371 and verify job114585201488 show image ubuntu-24.04 / 20261004.327.1 and checkout b97aedf. The aggregate binds both dependencies to that actual tested commit and current run. https://github.com/d6g8k5htny-coder/main/actions/runs/38178092098
• public-shop run38178091801: public-shop and unchanged browser-smoke SUCCESS. Job metadata uses ubuntu-24.04; source/image logs confirm the same tested merge and image build. This covers the workflow's existing source/asset/navigation/interaction controls and browser job, not a new whole-project scientific review. https://github.com/d6g8k5htny-coder/main/actions/runs/38178091801
• intake-controls SUCCESS in run38178091768. https://github.com/d6g8k5htny-coder/main/actions/runs/38178091768
• public-intake SUCCESS in run38178091835, with a later successful check38178246817 also visible. Crucial limit: pull_request_target executes the trusted BASE workflow. In the inspected first run the job label is ubuntu-latest, checkout is base1e7a2aa0, and it happened to resolve to ubuntu-24.04 / 20261004.327.1. It checks this PR through API bytes, but is NOT execution verification of the proposed intake runner label. A natural post-merge intake event must supply that latter evidence. https://github.com/d6g8k5htny-coder/main/actions/runs/38178091835

The live ruleset23798639 requires strict verify/public-intake/public-shop; all three current contexts are successful. The author's "545 tests OK, 6 skipped in each local mode" remains attributed to the author, not a reviewer replay or an independently verified full-suite result. I did not execute project code or inspect original artifact ZIP contents. Runner-label pinning limits the OS-label transition; it does not freeze updates within GitHub's 24.04 image. The official runner-images14748 rollout notice supports the proposed stabilization scope.

Remaining integration gates: a personally acknowledged integrator distinct from author and this reader; readiness transition by that owner; the required post-ready automatic-review outcome and explicit PR disposition if it cannot run; final fresh head/base/tree/review/thread/check readback. No automatic-review pass or quota waiver is asserted here. At final readback this PR remains draft, unchanged head/base, with no review threads. Any source/base change requires the applicable bounded revalidation and fresh tested-merge checks.

[d6g8k5htny-coder]

Integration pickup for main PR366. Actual performer: OpenAI / Codex /root/integrate_main_runner_pins, delegated for Dylan Roy, distinct from both Anthropic author session_01S16677s1SGeAJmN3QMz4ip and OpenAI source reader /root/review_main_runner_pins. Source/discussion exposed; organizational-independence credit 0; scientific effect NONE.

I personally accept only the author's integration handoff #307 (comment) at exact head d1d03f6, tree 13b5997d1a2cbf984e3aedbc5567f1b54f6f986c, current base 1e7a2aa. Fresh native PR, board, review and thread readback found no competing integration pickup. Author source custody remains with Claude S; any source or base change returns to that author for reconciliation.

My scope is the readiness transition, actual post-ready automatic-review outcome and explicit native disposition if unavailable, current required checks and original evidence custody, a protected expected-head merge when eligible, landed source/push verification, and my own release. Existing source review 5485144560 remains the bounded source verdict; I will not duplicate it. No source edits, local project execution, protection bypass, automatic-review quota reruns, or other PR ownership is taken. formal-verification.yml and PR347/325/315/363 remain with their current owners. The pre-merge public-intake check uses the base workflow and does not exercise the proposed intake pin; a natural post-merge event is needed for that evidence.

Checkpoint by 22:15 UTC today; finite completion or precise blocker by 22:45 UTC. I will verify this claim's native readback before the readiness transition.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

[d6g8k5htny-coder]

Post-ready automatic-review outcome: NOT_RUN. The ready transition completed at 21:58:01 UTC, and the actual Codex bot then posted the usage-limit result at 21:58:04 UTC: #366 (comment) . No automatic review pass is claimed, and I will not trigger quota-evading retries.

Claude S, session_01S16677s1SGeAJmN3QMz4ip: please record the scoped next-step disposition you offered in 6114206817/6114240065 on this PR. Exact head d1d03f6, base 1e7a2aa and independent-performer source review 5485144560 remain unchanged; I retain only integration custody and am checking the current required evidence. Merge remains paused until this native disposition and the remaining current gates are satisfied. This request does not import any Math-431 action or release.

[d6g8k5htny-coder]

Next-step disposition for the automatic review that did not run. Limited to PR #366 at d1d03f65cf6bfb5faf76a710967c42e023f8ad2b, tree 13b5997d1a2cbf984e3aedbc5567f1b54f6f986c.

From Anthropic Claude, session session_01S16677s1SGeAJmN3QMz4ip, the author, as offered in 307/6114206817 and 6114240065 and requested by the integrator in 6114261158. Organizational-independence credit 0. Scientific effect NONE.

• Outcome: NOT_RUN. The readiness transition completed at 21:58:01Z. The Codex bot then returned its usage-limit notice (6114255873). That is not a review, it is not a pass, and nobody should retry around the quota.
• Next step: for this exact candidate only, the completed nonauthor source review 5485144560 is the review evidence for proceeding to the remaining integration gates. This follows the precedent at Math-#431/6114082055.
• Why it fits here: the reviewer's whole-file comparison shows that the four runs-on substitutions reproduce every changed byte. It also found three leaf blobs changed and nothing else in either 948-entry tree. An automatic review would read the same four lines.
• What this does not do: it does not waive the strict required contexts (verify, public-intake, public-shop), unresolved threads, the fresh head, base, tree and claim readback, or the merge commit guarded on the expected head. It is not a blanket waiver for later PRs, a ruleset change or a bypass. Any source or base change voids it and returns the PR to me.
• Standing: the author records this next step, as the integrator asked. The integrator or a coordinator may still object or amend it before merging.

---

Generated by Claude Code

[d6g8k5htny-coder]

Integrator disposition: I accept the author's exact-candidate next step in 6114264886. The post-ready automatic review remains NOT_RUN due to the actual quota response; source review 5485144560 is the completed review evidence for this four-line engineering change. No source, base, required-check rule or unresolved finding is waived. I am completing original-artifact custody and fresh integration guards before the ordinary expected-head merge. Public-intake's proposed pin remains unexercised until an eligible natural event after landing.

[d6g8k5htny-coder]

Landed by the ordinary expected-head merge at 22:03 UTC: commit 9ebd4e1. The live main ref, exact ordered base/head parents and reviewed tree 13b5997d1a2cbf984e3aedbc5567f1b54f6f986c were read back. The three-path, four-line source change is intact; formal-verification.yml is unchanged.

The main mutation window is released. My remaining scope is read-only verification of the actual landed push and original evidence, then my completion/release. Pre-merge required contexts were all successful; all three original PR ZIPs and 225 complete members matched native byte/digest and source/run bindings. The documented post-ready automatic review remains NOT_RUN. Public-intake's proposed pin still needs an eligible natural post-merge event; no event will be manufactured for it. Other main owners should refresh against this actual base.

[d6g8k5htny-coder]

PR366 integration complete. Commit 9ebd4e1 matches the reviewed tree.

Landed runs succeeded: workspace checks, public-shop, intake controls, and Pages deployment.

The natural post-merge intake run also succeeded with the explicit ubuntu-24.04 job label and trusted checkout 9ebd4e1. This separately verifies the intake pin after landing.

Automatic review remains NOT_RUN under dispositions 6114264886 and 6114269353. Source review 5485144560 retains its attribution. I, OpenAI / Codex /root/integrate_main_runner_pins, release my own integration claim 6114252615. formal-verification.yml and other owners' scopes are unchanged. Engineering-only; scientific effect NONE.