review-swarm

To run a multi-agent code review on your git diff, use this skill to analyze your current changes and receive a prioritized summary of behavioral regressions and security vulnerabilities. It operates read-only across diverse codebases.

Parallel multi-agent code review evaluates git diffs simultaneously from four perspectives: intent/regression, security/privacy, performance/reliability, and contracts/coverage. It synthesizes feedback from specialized sub-agents into a high-signal, actionable report.

Yes, automated git code review requires access to local project documentation and git history to maintain context. This ensures the analysis aligns with existing architectural constraints and accurately identifies potential regressions.

The best way to catch security vulnerabilities and performance bottlenecks before a pull request is orchestrating a parallel multi-perspective analysis of your code changes. This approach filters out noise and provides a prioritized summary of risks.

Yes, you can use automated code review tools to detect behavioral regressions in diverse codebases. The multi-agent approach synthesizes parallel feedback from specialized sub-agents, ensuring high-signal detection across varied project architectures.

You should not use multi-agent code review if your project lacks git history or local documentation, as the analysis requires these to maintain context. Additionally, it operates strictly read-only, so it cannot automatically fix the identified issues.