[Nickfost]

A supported installation can retain an older last-known-good configuration with incompatible capacity or resource policy. These two guides explain explicit installed-baseline acceptance through the separately pinned helper from merged engine PR159, e61831c72787df57e76db3030ab7bb0b0986d369.

The instructions require authored managed-image policy, verified activation, exact installed checks and fresh state-aware health whose report agrees with its process exit. They use an explicit external-checkout command, explain refusal without repair or older-LKG fallback, preserve prior recovery evidence and holds, and require inspection of actual persisted results when failure follows LKG replacement. The immutable supporting guide is linked.

Documentation impact is limited to docs/UPDATING.md and docs/MANAGED-IMAGES.md. Ordinary runtime/policy840, executor/deployer689, configuration, schema, scripts, evidence, workflows, complete test inventories and historical release records are unchanged. Tracking: #41.

Complete exact-source acceptance passed on reviewed head/suite 5f53e2755afad161af0721ea6aff3c206b97e221, tree 91fcc19547521e9695617091d9c80e460b69f067, with actual upstream base/main 1bdef39a6932ef9ae2dfdddff34956a9ca8e1e80, independent policy840 and protected executor689. Every command retained its180-second limit.

• Upstream: all11 stages passed. Exported bundle SHA256 eb6e8c7c07050686686dec1a6b989f3bed0d5f3577d6f15ca1b129b3e9e87dff; report 30a9242c82ac4db1ece4d9e52f93c583aea82c719aa9263cad69187d7602eecf.
• Regenerated strict initialized adopter 1bfbd00a68de591c7a18dac9d0cf62527730939b, tree 0aca28d8df4e704de2712f0afdb318123ab7e577: all11 stages passed. Exported bundle SHA256 10800a8ed8b11278d7ed5843af5be2a43d382b7d8ea6871d4cac42c35451b35d; report 586f74f5b5f6cf2ab10a5fc27828f77cdb7431ddc93b24c3f02edaa890711573. Its genuine one-root/zero-base history applies only to this new fictional fixture.

Both17-member bundles were exported and independently checked against all16 indexed file hashes, all11 stage-log hashes, exact source/tree/base/suite/policy/executor identities and required inventories. Expected/actual skips are empty. Upstream policy109, updater16, release8 and harness50/6/18/14 passed. Adopter acceptance also retained complete public and mapped candidate corpora, upstream-probe counterpart accounting and the additional workflow corpus. All required IDs remain unchanged.

Current-head independent Codex review completed without findings. The earlier relative-command finding was fixed in documentation and resolved; all three unchanged documentation regressions pass. Local configuration, ten Markdown link/anchor checks, exact supporting-guide anchor, conventional commits/title, committed-history secret scan and whitespace passed.

Final cleanup/admission verified zero active validation processes/cgroups or running containers, unchanged daemon/four protected840 image tags, protected executor689 and the locked validation account. The retained operator key file is unchanged. No prune, tag removal, fleet installation, Actions runner or sudo grant was used. Optional hosted jobs remain skipped and supply no complete-suite evidence.

No new template release, private import, live helper use or storage/retention acceptance is claimed. Actual private import/full acceptance and authorized operations follow this accepted template revision.

---

Note

Low Risk
Markdown-only operator guidance; no scripts, schema, fleet configuration, or runtime behavior changes in this PR.

Overview
Documents how operators reconcile an already-installed approved runtime with a stale last-known-good (LKG) record that still describes incompatible capacity or resources.

docs/UPDATING.md adds Accepting a verified installed recovery baseline: use the separately pinned ci-fleet engine e61831c… and its immutable guide, preserve prior recovery evidence outside Git, then run remote-reconcile.sh --accept-installed from that checkout after repo checks pass. The text spells out lock/fetch/approval behavior, state-aware health (report must match exit), refusal to repair drift or fall back to the old LKG, post-failure inspection of persisted LKG, and explicit non-goals (storage admission, image deletion, legacy retirement).

docs/MANAGED-IMAGES.md adds a short note in Failure, removal and rollback that portable bindings plus an approved runtime do not auto-update older LKG, and points to the new UPDATING anchor before following supporting source order.

^{Reviewed by Cursor Bugbot for commit 5f53e27. Configure here.}

[Nickfost]

@codex review

Please review exact head 46d4db89b98389911ef5076f7f546c1348d8f90d against accepted main 1bdef39a6932ef9ae2dfdddff34956a9ca8e1e80. Only the two guides change. Compare the documented --accept-installed behavior with actual merged supporting source e61831c72787df57e76db3030ab7bb0b0986d369, especially independent command-source versus runtime/manager pins, activated managed-image prerequisite, locked installed commit/approval/checks, state-aware health report/exit consistency, and failure after LKG replacement. Guidance must preserve prior recovery evidence and holds, reject drift without repair or older-LKG fallback, and grant no cleanup or host authority. Ordinary policy840, executor/deployer689, scripts, inventories, workflows and historical release facts remain unchanged. Local documentation/configuration/history-secret/convention checks passed. Both full final-source upstream/initialized-adopter suites remain required on the explicitly released isolated VM before merge.

[chatgpt-codex-connector]

Codex Review Summary

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

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-31T22:14:19.867453Z	5f53e27	Draft marked ready

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: 46d4db89b9

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".

[Nickfost]

@codex review

Please review exact successor 5f53e2755afad161af0721ea6aff3c206b97e221, tree 91fcc19547521e9695617091d9c80e460b69f067, against unchanged accepted base1bdef. The only successor change makes the external supporting-checkout command explicit. The reviewed46d relative spelling reproduced a failure in the existing path-aware documented-artifact test; all three unchanged documentation cases now pass. The finding was fixed and its thread resolved. No checker/test/inventory change, pin change, engine change or deadline change occurred. Both final-source full VM suites remain pending; no suite ran on46d. Supporting helper remains actual merged e618, separate from ordinary policy/runtime840 and executor/deployer689. Please assess the final two-guide diff and external invocation, preserving state-aware health, post-LKG-replacement failure/hold requirements and no repair/older-LKG fallback.

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Hooray!

Reviewed commit: 5f53e2755a

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".

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: 29073f3b-f234-4860-9e59-72f250e0befa)