claude-commands

TAPE systematic debugging methodology: Think, Analyze, Plan, Execute. Use when: debugging errors, investigating failures, fixing bugs, diagnosing unexpected behavior. NOT for: code review, refactoring, performance profiling unrelated to a bug.

Novelty classification framework for extracting genuinely new knowledge from documentation. Use when: processing docs with /learn, extracting insights from large codebases, batch knowledge extraction. NOT for: summarizing documentation, generating training data, producing general overviews.

Codebase audit: security, debt, quality, prioritized findings, parallel scout architecture. Use when: auditing a codebase for security vulnerabilities, technical debt, quality issues, or discovering feature opportunities (roadmap mode). NOT for: single-file code review, formatting, or documentation writing.

Language-aware reusability-first refactoring with DRY optimization: project type detection, language rules, focus areas, duplication extraction, dead code removal, version sprawl elimination, and workspace cleanup. Use when: refactoring code for reusability, applying language-specific patterns, scanning for code smells, DRY extraction, eliminating duplication, cleaning workspace, removing version sprawl. NOT for: debugging, production audits, or adding new features.

Feature lifecycle: requirements, discovery, design approval, quality gates, production. Use when: planning feature lifecycle from requirements through design approval to production; crystallizing vague requirements into actionable specs. NOT for: bug fixes, refactoring, or code review without new feature context.

Notes to specs: DAWN structure, verification matrices, rollback plans, architecture decisions. Use when: converting rough notes, Slack dumps, or bullet points into structured developer specs; writing bug reports, feature specs, CLI tool specs, setup guides, or documentation. NOT for: code generation, API reference docs, changelog automation, or inline code comments.