Add a complexity-accretion tripwire - #152
Conversation
Work intentphase: completed
summary: "Complexity tripwire guidance merged in PR #152."
touch-points:
- "AGENTS.md"
- ".agents/skills/babysit/SKILL.md"
blocked-by: []
may-block: []
decisions:
- "AGENTS.md owns the canonical classification and user-pause rule"
- "babysit applies the canonical tripwire before any later review finding and pauses when it fires"
- "the workflow guard does not duplicate the canonical predicate"
validation:
- "npm run check"
- "npm test: 712/712"
- "focused agent-guidance tests: 3/3 after guidance edits"
- "skill quick validation: valid"
- "GitHub CI: Verify (Node 24) passed at exact head fb65db84"
- "Codex: explicit no-major-issues result at exact head fb65db84"
mergeCommit: "bf096db3de6bd73394d8cbe1d92ef2f844c1feaa"
branch: "agent/add-complexity-tripwire"
thread: "start-issue-5555c07af61c04f1b806de5c" |
![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: ba90012ed8
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 |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 4515f8a49b
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 |
|
Codex Review: Didn't find any major issues. What shall we delve into next? Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
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". |
Promote the approved babysit rewrite into the canonical skill locations and update the structural guidance test to follow its explicit merge-reference disclosure. Follow-up to #152.
* Add complexity tripwire * Stop unbounded babysit review loops * Bound tripwire history to current work * Delegate babysit stop to canonical tripwire
1 participant
What changed
The general tripwire pauses a defensive line of work when:
On a trip, the agent must report reachability, actor/interleaving, consequence, recovery, accumulated complexity, and the smallest alternative. User direction and a finite completion test are required before resuming.
Why
PR #141 showed that exact-head automated review can keep generating locally valid examples for an unbounded promise. Existing guidance required terminal-clean review but supplied no concrete stop condition. The result was a one-way complexity ratchet.
The general rule explicitly outranks terminal-clean babysitting while the concern is unresolved. The workflow-local guard places that decision at the point where babysitting would otherwise restart the review loop without duplicating the predicate.
Scope
This PR changes only
AGENTS.mdand.agents/skills/babysit/SKILL.md. It adds no framework, persistence, evaluation corpus, or application behavior.Validation
git diff --checknpm run checknpm test— 712/712 passingquick_validate.py— valid