review-changes

Automated code review dispatches parallel agents to scan pending git diff outputs for security vulnerabilities and performance issues. It identifies hardcoded secrets and bottlenecks, aggregating findings into prioritized P1-P3 categories before commit validation.

Checking architectural adherence involves scanning code diffs to detect dependency flow violations and anti-patterns. Dedicated review agents analyze staged changes to ensure pattern compliance, generating a structured summary of blockers, warnings, and notes.

Parallel review agents can validate FFI null pointers and unsafe blocks by scanning diff outputs. The security agent specifically targets these high-risk areas alongside hardcoded secrets, ensuring unsafe code segments are flagged before integration.

Multi-agent pre-commit validation is needed when completing features or modifying multiple modules, especially those involving FFI. It ensures code simplicity and security by running five specialized agents before you commit significant changes.

Automated code review operates without external dependencies to detect hardcoded secrets. It uses internal scripts and references to scan diff outputs directly, identifying security vulnerabilities and complexity issues independently.