What problem does it solve? Ad-hoc code written outside a formal plan-and-implement flow often ships without review, and blindly accepting every AI reviewer suggestion can degrade quality. This Skill runs multiple independent reviewers in parallel, critically evaluates each finding against project conventions, and applies only the fixes that are actually valid. ## Core Features & Use Cases - Parallel multi-reviewer execution: Launches Claude subagent, CodeRabbit, GitHub Copilot CLI, and Gemini CLI reviewers simultaneously against diffs, staged changes, branches, or pull requests. - Critical triage of findings: Merges duplicate findings and evaluates each for technical correctness, alignment with project design policies, intentional design decisions, fix risk, and cost-benefit before acting. - Automated fixing and PR integration: Applies approved fixes, verifies tests still pass, and optionally posts an inline review comment to the GitHub PR with an ai-review label after user confirmation. - Use Case: After quickly hacking together a feature without going through the normal planning flow, run the skill to have four independent reviewers examine the diff, then automatically fix only the findings judged worthwhile. ## Quick Start Ask the assistant to self-review the current uncommitted changes and fix any valid issues it finds.
