Restore durable Codex goal prompt headroom#148
Conversation
|
Warning Review limit reached
Next review available in: 32 minutes Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (6)
WalkthroughCodex prompt templates now require 300 characters of headroom and can split route groups earlier. GMCC-v1 completion semantics were standardized across workflow and skill prompts, with expanded checker and test coverage for alignment, readiness, and auto-merge closeout behavior. ChangesCodex sizing and completion contracts
Estimated code review effort: 4 (Complex) | ~45 minutes Possibly related PRs
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![claude[bot]](https://avatars.githubusercontent.com/in/1236702?s=60&v=4){kind=small}
|
Review: Restore durable Codex goal prompt headroom (#148) Overview This PR raises the Codex goal-prompt headroom floor from 100 to 300 characters, proactively splits route groups when a valid-but-tight prompt drops below that floor, and replaces the verbose, repeated "Goal Mode Completion Contract" paragraph with a compact, versioned This repo relies heavily on natural-language prompt text to drive autonomous agent behavior, so the review focused on semantic precision of that text (since ambiguity there is the equivalent of a logic bug) as well as the Ruby validation scripts. Findings
Code quality / test coverage
Security / performance
Overall this is a well-tested, self-contained change with good regression coverage for the character-budget mechanics; the main substantive concern is the ambiguous scope of the blocker exception in the new compact prompt line, since that is the text agents will actually act on. |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
ReviewReviewed the diff ( What this PR does
Correctness
Minor finding (posted inline)
Security / performanceNo security-relevant surface changed (no new external input handling, no code execution paths beyond the existing internal validation tooling). Performance is a non-issue — all changes are string-length/content checks over small, fixed markdown templates. Test coverageGood — the PR adds targeted regression tests for the exact two semantic gaps called out in its own description, plus mutation-style byte-parity assertions that will fail loudly on any future drift between the four |
|
Review: Restore durable Codex goal prompt headroom (#148) Overview What I checked
Code quality / correctness
Security / performance Test coverage Overall this looks like a solid, carefully self-verified change. The two inline items are suggestions/questions rather than blockers. |
Review SummaryWhat this PR does: Replaces the verbose, repeated "Goal Mode Completion Contract:" prose block (previously pasted byte-for-byte into 4 goal-prompt locations) with a compact, versioned Code quality / correctness
Minor observations (posted inline, non-blocking)
Security / performanceNo security-relevant surface here — this PR only edits Markdown prompt specs and local Ruby validation/test scripts that read repo files; nothing touches network input, shell execution with untrusted data, or secrets. Performance is a non-issue (small, one-shot string/regex checks over local files run in CI). Overall this is a well-scoped, heavily self-validated change with good before/after evidence in the PR description. I didn't find any correctness bugs blocking merge. |
QA Evidence
|
ReviewThis PR replaces the verbose, repeated Verification performed
Minor observations (non-blocking)
OverallNo correctness bugs or security issues found. The change is a well-scoped, heavily self-validating documentation/prompt-engineering refactor (six files, all cross-checked by an expanded Ruby contract/size test suite), and the described "prompt pressure replay" numbers are consistent with the code paths that measure them. |
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@skills/plan-pr-batch/SKILL.md`:
- Around line 322-328: Adjust the indentation of the two sizing bullet items in
the goal-prompt sizing section so they use the same two-space indentation as
their sibling list items. Preserve their text, ordering, and surrounding
fence/counting instructions.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: 7835b423-7ddb-4f7d-a5a7-f6bcb80d44fc
📒 Files selected for processing (6)
skills/plan-pr-batch/SKILL.mdskills/plan-pr-batch/scripts/check_goal_prompt_size.rbskills/pr-batch/SKILL.mdskills/pr-batch/bin/goal-completion-contract-test.rbskills/triage/SKILL.mdworkflows/pr-processing.md
ReviewOverview: This PR reworks the "Goal Mode Completion Contract" that's embedded in generated Codex/Claude goal prompts, replacing the verbose inline copy with a compact, versioned Verification: I confirmed the Findings
Overall this is a well-scoped, well-tested doc/prompt-engineering change. The only actionable item is the stray indentation on two bullets. |
![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: 7e53a949e2
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".
Review SummaryThis PR reworks the Goal Mode Completion Contract into a versioned, terse Overview
Verification performed
Finding
Other observations (non-blocking)
No security or performance concerns — this PR only touches Markdown documentation/prompt templates and Ruby validation/test scripts with no runtime/production code paths. |
Review: Restore durable Codex goal prompt headroomOverviewThis PR compacts the repeated, verbose "Goal Mode Completion Contract" prose into a versioned, self-contained I verified this statically (grep/manual cross-referencing across all 6 changed files) since the sandbox here blocks executing
Code quality observations (non-blocking)
Correctness / semanticsThe compact Security / performanceNo security-relevant code paths (docs + Ruby validation scripts only, no runtime app code). Performance impact is negligible — this only affects prompt-generation guidance and CI-time validation script execution. Test coverageTest coverage is thorough for a prompt/documentation change: mutation tests ( Overall: solid, well-tested change with strong internal consistency. No blocking issues found. |
ReviewThis PR compresses the verbose Verification performed (static, since this sandbox blocks executing
Bugs / security / performance: none found. This PR only touches Markdown documentation and a Ruby text-validation script with no user input or external execution surface, so there's no meaningful security or performance angle here — the main risk category is textual drift/inconsistency between the multiple copies of the contract line, and the new tests specifically close that gap (byte-for-byte alignment checks, mutation tests for the legacy closeout sentence and the configured-reviewer omission). I wasn't able to execute No inline comments — I didn't find any specific lines that need to change. |
QA Evidence
|
1 participant
Closes #123
Summary
GMCC-v1line while retaining the full canonical expansion inworkflows/pr-processing.mdPrompt pressure replay
Validation
MISE_CACHE_DIR=/tmp/aw-d-i123-mise-cache RUBOCOP_CACHE_ROOT=/private/tmp/aw-d-i123-rubocop-cache bin/validateReview history
Summary by CodeRabbit
Documentation
Tests
Observed model and effort provenance
Audited from Codex session
turn_contextrecords, worker handoffs, and commit timestamps. These observed routes supersede requested-route labels in earlier handoffs/comments. Git author metadata remainsjustin808; the table records AI content provenance separately.gpt-5.6-terra/highgpt-5.6-sol/highgpt-5.6-sol/xhighgpt-5.6-sol/highgpt-5.6-sol/xhighgpt-5.6-sol/highgpt-5.6-sol/highgpt-5.6-sol/highCommit provenance
164404dRestore durable Codex prompt headroomgpt-5.6-sol/high.d40be30Clarify compact goal closeout semanticsgpt-5.6-sol/high.a3d6a4bAlign compact goal fallback guidancegpt-5.6-sol/high.801f903Align compact review gate semanticsgpt-5.6-sol/high.7e53a94Fix prompt sizing list indentationgpt-5.6-sol/high.bc55840Fix Output list nestinggpt-5.6-sol/high.da0bf62Preserve configured reviewer gate in GMCCgpt-5.6-sol/high.3656708Align canonical auto-merge closeout semanticsgpt-5.6-sol/high.