joelyoung01/joelyoung01.personal-agent-dev
Personal agent development skills: pre-ship code review, git work-tree cleanup, log diagnosis, and token-efficient caveman communication.
Ultra-compressed communication mode. Cuts token usage ~75% by speaking like caveman while keeping full technical accuracy. Supports intensity levels: lite, full (default), ultra, wenyan-lite, wenyan-full, wenyan-ultra. Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman. Also auto-triggers when token efficiency is requested.
Analyze all uncommitted changes in the git working tree (staged, unstaged, and untracked files), organize them into logical commits, and create those commits without asking for approval. Use when the working tree has accumulated many uncommitted changes, the user wants to clean up their work tree, organize changes into commits, or asks to tidy up uncommitted work. Prefer delegating the full workflow to a subagent to preserve parent context.
Runs a structured diagnosis workflow when the user provides logs that contain an error or unexpected behavior. States one or more hypotheses for the failure, investigates relevant files (schema, config, source), plans and implements a fix for the primary hypothesis, and verifies the fix. Use when the user shares terminal output with an error, pastes failed command output, provides application/server logs, shares build or CI output, or asks to diagnose any log-based failure.
Load when the user has finished implementing changes and wants a rigorous pre-ship review — bugs, security gaps, architecture missteps, dead code, redundancy, and fit with existing project patterns. Also load when explicitly asked to review changes before merge or ship.