Create new agent skills with proper structure, progressive disclosure, and bundled resources. Use when user wants to create, write, or build a new skill.
Execute PLAN.md files using intelligent strategies based on checkpoint types. Use when user says 'execute', 'run plan', 'build it', or wants to progress from PLAN to SUMMARY.
Create executable prompts for Claude Code. Use when user asks to create a prompt, generate a prompt for a task, or write a prompt for another agent.
Execute prompts from .principled/prompts/ via delegated sub-tasks. Use when user asks to run, execute, or launch prompts.
Review all git changes — triage each as ship, gitignore, or discard, then commit, push, and sync all worktrees.