Conversation
The effort whitelist was a stale copy of the Codex enum, so pinning cross_model_effort to none or max skipped the peer as incompatible. Keep fail-closed for values the route cannot honor. Rebased onto current main via Git Data API (OAuth lacks workflow scope for pushing main CI history).
![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: 68fbadb210
ℹ️ About Codex in GitHub
Codex has been enabled to automatically 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 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
302090f to
4e79025
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 4e790258c9
ℹ️ About Codex in GitHub
Codex has been enabled to automatically 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 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
Keep the Codex none/max effort change from dropping +x on cross-model-adversarial-review.sh and cross-model-doc-review.sh.
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 34d9a3502c
ℹ️ About Codex in GitHub
Codex has been enabled to automatically 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 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
| expect(readFileSync(path.join(repoRoot, "docs/guides/configuration.md"), "utf8")).toContain( | ||
| "codex `none`..`max`", | ||
| ) | ||
| const yamlNeedle = "codex none|minimal|low|\n# medium|high|xhigh|max" |
There was a problem hiding this comment.
This assertion includes the literal newline and comment prefix, so a semantically neutral reflow of the YAML comment will fail the full test suite even when the documented enum remains unchanged. Normalize comment whitespace or extract and compare the enum values so the guard pins the protocol rather than incidental formatting.
AGENTS.md reference: AGENTS.md:L216-L220
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Fixed in cebfd4e. The yaml check now strips comment prefixes and whitespace before comparing, so it pins the enum values rather than the comment's line wrapping. Verified both directions locally: a reflowed comment passes, a changed value list still fails.
…t wrapping Codex review on EveryInc#1567: the yaml needle embedded a literal newline and comment prefix, so a semantically neutral reflow of the comment would fail the suite. Strip comment prefixes and whitespace before comparing so the guard pins the enum itself.
|
Thanks for taking this one. Something landed after you wrote it, and the PR doesn't cover it yet. #1634 merged on Sep 10 and shipped in 3.25.0. It copied the codex allowlist into codex:minimal|codex:low|codex:medium|codex:high|codex:xhigh) ;;Its test pins the old belief too, in rejected("codex", "max") // codex tops out at xhigh
rejected("codex", "none")The drift guard here reads the two review scripts and nothing else, so CI stays green while the copies disagree. Once this merges, I re-ran the junk-value probe on codex-cli 0.154.0 today. The API still lists Could you fold |
3 participants
Summary
cross_model_effortrejected Codex CLI valuesnoneandmaxeven though those are validmodel_reasoning_effortenums (Codex CLI 0.150.1). Fail-closed is still correct for values the route cannot honor; the whitelist was a stale copy of the provider enum.This PR adds
codex:noneandcodex:maxto both review workers (validate_effort_overridestays byte-identical), updates the comment/docs/config copies of that list, and pins the seven-value enum in tests so the next CLI addition cannot silently drift.Editorial default remains
xhigh.ce-work/ce-povhardcodedhighis unchanged.Validation
bun test tests/skills/ce-code-review-cross-model-routes.test.ts tests/skills/ce-setup-check-health.test.ts— 161 passbun run release:validate— in syncCloses #1565
Security Disclosure
No security-relevant changes.
Agent Disclosure