Review Issue

Goal

Provide an independent quality gate between implementation and production. This skill runs in a fresh Claude session (not the one that wrote the code) and orchestrates subagents to audit the diff, fix issues, and re-audit in a loop until the code is clean — then ships it.

The orchestrator itself stays thin. Heavy work (reading diffs, auditing code, making fixes) is delegated to subagents so the main context is protected and the audit loop can run as many times as needed without degradation.

When to Use

• After /plan-issue + /implement-issue completes in a separate session
• You are in a new Claude session (fresh context, no implementation bias)
• The feature branch has committed or uncommitted changes ready for review
• You want the only quality gate before prod to be rigorous

Prerequisites