Promote start-issue skill rewrite - #154
Conversation
phase: completed
summary: "PR #154 merged into main as d00d5e9673aff0fcd66379c4a0530576e9e44a84 after exact-head CI passed; the approved stop condition ended automated review after the final one-word correction."
touch-points:
- ".agents/skills/start-issue/"
- ".agents/skills/babysit/"
- "evals/start-issue/cases.json"
- "test/start-issue-evals.test.ts"
done-when: "PR #154 is merged into main after exact-head CI passes and the approved stop condition is recorded."
excludes:
- ".agents/skills/eli5-html-doc-rewrite/"
- "application code and runtime behavior"
- "new tracker, persistence, trust-allowlist, owner-token, claim-election, exhaustive Project-scan, or further automated-review machinery" |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: c17f0b2dc1
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".
|
@codex review Current head: |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 0d7b26f4ec
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".
|
@codex review Current head: |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: f7cf3afdc1
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".
|
@codex review Current head: |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: ab0b38a0e5
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".
|
@codex review Final authorized head: |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 820802a927
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".
1 participant
What changed
start-issuerewrite into the canonical skilldone-when/excludesslice boundaries for implementation and babysitfolddisposition, place its tripwire before triage actions, and complete merged PR claimsWhy
The previous workflow accumulated ownership tokens, stable-set loops, exhaustive unmarked-item reconstruction, and repeated scans around recoverable coordination races in a single-user project. The rewrite keeps the useful human-visible tracker and claim outcomes while making ordinary runs finite and understandable.
The babysit follow-up closes the cross-skill lifecycle seam and gives valid in-slice cleanup findings a bounded home without opening a review round solely for polish.
This is a standalone selected follow-up from the process analysis archived in #151. It does not depend on #151 merging.
Impact
Agent workflow guidance and its eval corpus only. No application code, runtime behavior, project configuration, trust allowlist, owner token, or compatibility layer.
Validation
quick_validate.pyfor both skillsnpm run ci:local— 714/714 tests plus smoke passinggit diff --check