Use when reviewing pull requests, reviewing your own code before committing, or when asked to review code changes. Provides structured analysis with severity categorization, security focus, and optional worktree isolation for safe PR review.
Use when creating, restructuring, or scaling a CLAUDE.md file. Provides tiered loading patterns, organization templates, and token measurement guidance. Prevents the "unmanageable blob" problem.
Use when encountering any bug, test failure, or unexpected behavior. Provides a structured 4-step cycle and ACH methodology for complex bugs. Replaces random guessing with evidence-based investigation.
Use when facing any non-trivial task to decide whether to handle it yourself or delegate to a subagent. Provides score-based routing, model selection, and structured prompt templates for effective delegation.
Use when you have a multi-step task, before touching code. Creates bite-sized implementation plans with TDD emphasis, then executes them in batches with review checkpoints. Prevents the "just start coding" trap.