claude-code-skills

Capture this session's repeatable process into a reusable skill. Call at end of the process you want to capture.

Audit Agent/Task model routing — reads the dispatch log written by the model-routing PreToolUse hook and breaks down spend by model, subagent type, and project. Use when the user asks 'how's the model routing', 'audit my model usage', 'check the dispatch log', 'is the routing hook working', '/model-audit', or before drafting any public-facing post about Claude Code usage. Requires the model-routing hook to be installed first (see hooks/model-routing in this repo).

End-of-session context sync. Reviews what was learned in the conversation and proposes updates to knowledge/, people/, and projects/ files. Use when the user says /sync-context, 'sync context', 'update context', or at the end of a session in a project that uses the compounding-context structure.

Bootstrap progressive disclosure context architecture in any project. Turns CLAUDE.md into a router with a task-based routing table, consolidates context into a single knowledge directory, and cleans MEMORY.md into a pure index. Use when starting a new project, when CLAUDE.md has 'always read' directives, when context feels bloated, or when the user says /init-context.

Optimizes prompts using full GEPA methodology (Genetic-Pareto Evolution). Use when user wants to improve a prompt's accuracy on test cases, mentions "optimize prompt", "improve prompt", or has examples of desired input/output pairs. Implements Pareto frontier selection, trace-based reflection, and crossover mutations.

Personal tutor that teaches any topic. Use when user says "teach me", "learn about", "quiz me", "/skill-tutor", or wants to understand a new concept deeply. Creates personalized tutorials using user's actual projects, tracks learning progress, uses spaced repetition.