reviewing-code

You can automate code review by asking the agent to review a specific PR number or current changes. It detects the scope, runs git/gh diff commands, and returns a structured findings report covering bugs, security risks, and missing tests.

Structured AI code review systematically analyzes diffs and changed files to find bugs, security risks, and style violations. It enforces a checklist-driven protocol and returns a standardized report with prioritized Must Fix, Should Fix, and Consider findings.

Yes, you can review staged, unstaged, or specific files directly without a pull request. The Skill detects your current git context, reads the diffs, and generates a structured checklist-driven findings report for your working tree changes.

Yes, the code review process explicitly checks for missing tests and documentation drift. It analyzes changed files against a checklist to surface these issues alongside bugs, security risks, and style violations in the final report.

Review large diffs efficiently by spawning a fresh-context subagent to run git/gh diff commands and file reads. This keeps large diffs out of the main context for token efficiency while returning a strict structured findings report.

Yes, the code review can align with an existing feature plan file. It looks up associated plan files using F-XXXX patterns and compares the implementation against stated deliverables when generating the findings report.