docs: note that a cancelled ci-status run blocks the merge until a new commit

[kyle-sexton]

No related issue: operator gotcha from the automation-lanes program

Summary

Records in the README Contract section that a cancelled ci-status run keeps the merge blocked.

Fix

Adds one bullet: every same-named required check run on a PR head must pass; a cancelled run leaves a failed check beside a later successful one; re-running does not clear it, a new commit does.

Verification

markdownlint-cli2 README.md: 0 issues.

Related

Observed in the automation-lanes program (handoff constraint h17).

🤖 Generated with Claude Code

https://claude.ai/code/session_01EFqTQRnCGywRxhv4nw4o4V

[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 2026-10-10T07:36:36.314111Z	f166805	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.

[github-actions]

claude-security-review has reviewed this pull request through c434700; a later push is reviewed from there.

[claude]

Claude finished @kyle-sexton's task in 3m 43s —— View job

---

Code review — /review:code-review

• Gather context (diff vs origin/main, surrounding README sections, related pr-require-checks scripts)
• Understand the request (CI code-review lane for this doc-only PR)
• Review changed lines for correctness/maintainability
• Post findings

Scope: README.md (+4/-0) — adds one Contract bullet:
L57-L60.

No findings. This is a narrow, accurate documentation addition:

• The new bullet is consistent with the adjacent Contract bullets (style, backtick usage, bold-lead-in convention) and doesn't contradict them — the preceding bullet (L50-L56) describes the gateway job's own cancellation handling within a single run; this bullet describes a separate, external GitHub behavior across multiple check runs sharing a name on the same SHA.
• I checked .github/actions/pr-require-checks/aggregate-results/run.sh's rerun-contract-only-siblings mechanism for a possible contradiction (it re-runs a failed contract-only run so "its red check run is replaced by one that reads the success"). That's a mechanistically different scenario — a completed failure run being re-run — from the cancelled-run-beside-a-later-success case this bullet documents, so I don't think it undercuts the claim, but flagging the distinction in case the author wants to tighten the wording.
• MD013 (line length) is disabled in .markdownlint-cli2.jsonc, so no lint risk from line wrapping.
• Author-claimed, unverified by me: the PR body's markdownlint-cli2 README.md: 0 issues — I could not independently run markdownlint-cli2 in this sandbox (network/npx tool calls require approval that wasn't available here), so I'm not confirming that result myself.

Nothing to block or request changes on.
· branch

🤖 Generated with Claude Code

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: f166805dda

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".

[github-actions]

claude-review has reviewed this pull request through c434700; a later push is reviewed from there.

[claude]

Claude finished @kyle-sexton's task in 2m 47s —— View job

---

Code review — /review:code-review (incremental)

• Gather context (incremental diff since f166805, README surrounding sections)
• Understand the request (CI code-review lane, incremental scope: re-review the README.md edit made in response to the Codex P2 comment)
• Review changed lines for correctness/maintainability
• Post findings

Scope: incremental diff only — README.md Contract bullet, L57-L63. This commit narrows the bullet in direct response to chatgpt-codex-connector's P2 finding on the prior revision (universal "every run must pass" claim vs. the documented carry-forward supersession).

No findings. The rewrite holds up:

• It no longer states a universal rule. It now says a newer successful ci-status normally supersedes an older red one, pointing at record-pending/rerun-contract-only-siblings (L349, L357), and scopes the gotcha to a run cancelled by concurrency — no longer contradicting L377-L385's "the full run's later ci-status supersedes the red" claim.
• The scoped case is consistent with the mechanism described below it: rerun-contract-only-siblings sweeps "every failed run of the same workflow on the SHA" via rerun-failed-jobs (L357-L361) — a cancelled conclusion isn't a failure conclusion, so a cancelled run's check plausibly falls outside that sweep. The new text doesn't spell out this mechanism, but it doesn't claim one that contradicts it either.
• (see X below) mid-sentence matches existing README convention (e.g. L252, L329).
• Bold lead-in, backtick usage for ci-status, record-pending, rerun-contract-only-siblings all match sibling bullets.
• Author-claimed, unverified by me: markdownlint-cli2 README.md — I couldn't run it in this sandbox (network/npx tool calls require approval not available here), same limitation as the prior review of this PR.

Nothing to block or request changes on.
· branch

🤖 Generated with Claude Code