taches-principled

Apply systematic debugging methodology to investigate issues deeply

Simplify and refine recently modified code for clarity and maintainability

Create a handoff document for continuing work in a fresh session

Synthesizes multiple evaluation results into an actionable improvement plan. Use after grading and comparing skills to generate prioritized recommendations.

Reviews code for quality, security, and best practices. Invoke when user asks for code review, PR review, or code quality feedback.

Compares two skill versions to identify what changed and why it matters. Use when a skill was revised and you want to understand the delta.

Scores skill outputs against teaching-focused rubrics. Use when evaluating whether a skill teaches judgment or just prescribes steps.

Expert prompt engineer for Claude Code. Use when creating, optimizing, or executing reusable prompts for task automation. Invoke when user asks to 'create a prompt', 'write a prompt', 'make a prompt executable', or 'run a prompt'.

Creates draft task files in .specs/tasks/draft/ from user intent. Use when user says 'capture this task', 'create a task', 'add to backlog', or 'log this as something to do'.

A3-format problem analysis with root cause investigation and action plan. Use when user says 'analyze this problem', 'root cause', 'why is this happening', or 'figure out what's wrong'.

Auto-selects improvement method (code exploration, workflow mapping, or waste identification) for a given target. Use when user says 'analyze this', 'look into this code', 'what's wrong here', or 'find the problem'.

Multi-agent code review for PRs and local changes. Use when user says 'review this PR', 'check my changes', 'review the code', or 'audit this pull request'.

Refactors complex code to reduce cognitive load. Use when user says 'simplify this', 'clean up this code', 'reduce complexity', or 'this is too nested'.