feat: dual-model cross-validation in PM review stage#628
Open
lzy5818957 wants to merge 2 commits intogarrytan:mainfrom
Open
feat: dual-model cross-validation in PM review stage#628lzy5818957 wants to merge 2 commits intogarrytan:mainfrom
lzy5818957 wants to merge 2 commits intogarrytan:mainfrom
Conversation
Adds three new skills for autonomous project management: - /pm: Takes a GitHub issue URL and drives it through the full pipeline (intake → plan → build → review → ship) autonomously, only escalating to the user for taste decisions, critical failures, or blockers. - /status: CEO dashboard showing current state of all PM-managed sprints — what's in progress, current stage, bugs found, blockers, and key metrics. - /report: Sprint report with detailed per-sprint and cross-sprint analysis showing lines changed, bugs caught, test coverage delta, stage durations, decision audit, and team performance trends. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The PM's review phase now runs both Claude (/review) and Codex (/codex review) in parallel, then cross-validates findings to filter false positives: - Both agree → confirmed, fix it - One only, critical → ask the other model to verify before fixing - One only, non-critical → auto-dismiss (most common false positive) - Contradictory → escalate to user This prevents the common case where a single reviewer flags a false bug and the fix breaks working code. Sprint metrics now track per-model findings, confirmed count, and false positives filtered. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
/review) and Codex (/codex review) in parallelDepends on: #626 (PM layer skills)
Problem
Single-model code review frequently flags false bugs. When the PM auto-fixes these, it breaks working code. This is the most common failure mode in autonomous pipelines.
Cross-validation logic
Test plan
bun test— skill validation passesbun run gen:skill-docs— pm/SKILL.md regenerates correctly/pm <issue-url>— review stage runs both models and shows cross-validation table🤖 Generated with Claude Code