What problem does it solve? Reviewing a pull request thoroughly requires repeated audit-and-fix rounds, and driving each round manually is slow and error-prone. This Skill automates the full cycle: it gates the diff against CODE_RULES, spawns a clean-room audit agent each loop, dispatches a fix agent for the findings, posts per-loop GitHub reviews, and stops only when the PR converges, hits the 20-loop cap, or gets stuck. ## Core Features & Use Cases - Clean-room audit loops: Each loop spawns a fresh code-quality-agent (opus) with no inherited context, auditing all 17 categories (A–Q) from API contracts to concurrency hazards, with findings posted as one anchored GitHub PR review per loop. - Targeted automated fixes: A fresh clean-coder worker receives only the latest audit's findings, applies narrow fixes, runs tests, commits and pushes, then replies to and resolves each review thread. - Deterministic quality gate: A code_rules_gate.py pre-audit check must pass before every audit, with up to 5 standards-fix rounds to clear violations. - Use Case: Before requesting human review on a large PR, run /bugteam to autonomously iterate audit and fix loops until the diff reports 0 P0/P1/P2 findings, with a full loop log and final report on exit. ## Quick Start Run /bugteam on the current branch's open pull request to start the automated audit-fix loop until it converges.
