mp-prd-review

To review a PRD implementation, you can aggregate merged PRs and sub-issues to compute a diff against the merge base, then run parallel scans for code quality, security, and architecture gaps. This identifies cross-PR patterns and classifies findings by severity.

A PRD phase-end quality gate is a comprehensive review process that validates spec coverage, checks cross-PR patterns, and identifies security or documentation gaps. You need it during PRD wrap-ups to systematically reduce risk before closing a development phase.

You generate a PHASE_END review document by collecting PRD sub-issues and merged PRs, running parallel sub-analyses on the aggregate diff, and synthesizing findings into Critical, Important, and Minor categories before waiting for user approval.

Yes, you can run automated spec coverage validation by pulling PRD sub-issues and their merged PRs, computing an aggregate diff against the correct merge base, and scanning for unresolved items to ensure the implementation aligns with the specification.

The best way to identify dead code and cleanup needs is by running targeted parallel sub-analyses on the aggregate diff of merged PRs. This scans for architecture gaps, dead code, and cleanup requirements, deduplicating findings into a prioritized review document.

Yes, the PRD review process requires a user-confirmed Human-in-the-Loop (HITL) gate before executing fixes. After generating the PHASE_END review document and classifying findings by severity, the system waits for your explicit approval to proceed.