codebase-review-swarm

What problem does it solve?

This Skill eliminates the risk of unverified, hallucination-prone AI code reviews by enforcing strict quote-grounded evidence, mandatory coverage closure, and multi-stage reviewer/critic validation for all findings, ensuring only verified, actionable results are produced.

Core Features & Use Cases

• Multi-track audit support: Run focused reviews for security, QA, accessibility, performance, AI-slop, supply chain, or enhancement opportunities, or a complete integrated review covering all domains.
• Non-diluting depth guarantee: Selected tracks receive the same exhaustive depth whether run alone or as part of a multi-track review, with no sampling or shallow checks.
• Structured artifact output: Generates machine-readable review artifacts, coverage ledgers, and a final verified report without modifying any source files.
• Use Case: A team preparing a high-risk feature for production can run a focused security and QA audit to catch hidden defects, verify all claims against actual code, and receive a report with exact evidence for every finding.

Quick Start

Invoke the codebase-review-swarm skill at your repository root to run a full evidence-gated codebase review, select your desired audit tracks after the initial inventory phase, and receive a verified report with supporting artifacts without modifying any source files.