メニュー
Score skill fitness with GDI and auto-improve SKILL.md files using the gene pool (EvoMap) and compounding cycles (BotLearn). Use when user asks to improve skills, audit the skill ecosystem, run skill evolution, check skill health, or /skill-evolve.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Use when you need to generate or enhance user interface and user experience designs.
Manage Things 3 via the `things` CLI on macOS (add/update projects+todos via URL scheme; read/search/list from the local Things database). Use when a user asks Clawdbot to add a task to Things, list inbox/today/upcoming, search tasks, or inspect projects/areas/tags.
Collaborative thinking partner for exploring complex problems through questioning
Use when you need to access and apply superpowers within your Claude projects.
A universal self-improving agent that learns from ALL skill experiences. Uses multi-memory architecture (semantic + episodic + working) to continuously evolve the codebase. Auto-triggers on skill completion/error with hooks-based self-correction.
Use when you need to analyze your own reasoning process to improve accuracy.
| name | skill-evolve |
| description | Score skill fitness with GDI and auto-improve SKILL.md files using the gene pool (EvoMap) and compounding cycles (BotLearn). Use when user asks to improve skills, audit the skill ecosystem, run skill evolution, check skill health, or /skill-evolve. |
Full EvoMap + BotLearn self-evolution loop for the PAI skill ecosystem:
EvoMap GDI scoring — every skill gets a fitness score:
GDI = usage_frequency(35%) + recency(30%) + content_quality(20%) + trigger_clarity(15%)
Gene pool (compounding) — proven improvement strategies accumulate in _logs/gene_pool.json.
Each successful improvement extracts a gene. Future improvements try matching genes first
(free, deterministic) before AI generation. Confidence grows with each application.
BotLearn loop — every evolution cycle makes the system smarter and cheaper to run.
# View GDI fitness scores for all 130+ skills
python3 _scripts/skill_evolution.py fitness
# Show only high-priority skills (GDI < 0.3)
python3 _scripts/skill_evolution.py fitness --priority high
# Improve a skill (tries gene pool first, then AI fallback)
python3 _scripts/skill_evolution.py evolve --skill <name>
# Batch-improve all high-priority skills
python3 _scripts/skill_evolution.py evolve --all --priority high --confirm
# View the gene pool (proven strategies)
python3 _scripts/skill_evolution.py gene list
python3 _scripts/skill_evolution.py gene stats
# Find scripts/patterns that need new skills
python3 _scripts/skill_evolution.py synthesize
# Weekly fitness report (also runs automatically via /org-weekly)
python3 _scripts/skill_evolution.py report
skill_tracker_hook.py → _logs/skill_usage.json
↓
skill_evolution.py fitness (GDI scoring, no AI)
↓
skill_evolution.py evolve (gene pool first → AI fallback)
↓
extract gene → _logs/gene_pool.json (confidence compounds)
↓
next cycle: gene pool is smarter
_logs/skill_usage.json — usage events (auto-populated by UserPromptSubmit hook)_logs/gene_pool.json — compounding library of proven strategies_logs/skill_backups/{skill}/{ts}/SKILL.md — pre-improvement snapshots_logs/skill_evolution_applied.json — improvement historySources/Skill Fitness Report - YYYY-MM-DD.md — weekly report