Anthropic published a study in February classifying 11 observable collaboration behaviors across 9,830 Claude conversations. Reading it, I realized I had no way to know which of those 11 behaviors appeared in my own sessions — or which ones I'd never touched in months of daily Claude Code use.
skill-tree runs that same classification against your local session history. It extracts your user messages, sends them to a remote classifier (Claude Haiku on Fly.io), maps them onto the same 11 behaviors from Dakan & Feller's 4D AI Fluency Framework, then assigns one of seven archetype cards — rendered as tarot cards with curated museum art — and picks one behavior you haven't tried as a growth quest for your next session.
The part I found most useful: seeing the gaps. The framework organizes behaviors across three axes visible in chat logs — Description, Discernment, Delegation. I was heavy on Description and light on Delegation. Obvious in retrospect, invisible before.
Live example of what the output looks like: skill-tree-ai.fly.dev/fixture/illuminator
The 7-step orchestration (find session files → extract messages → classify → assign archetype → synthesize narrative → render → return URL) runs in 30–60 seconds and returns a stable URL. The growth quest persists across sessions via a SessionStart hook, stored at ~/.skill-tree/.
Install in Claude Code:
claude plugin marketplace add robertnowell/ai-fluency-skill-cards
claude plugin install skill-tree-ai@ai-fluency-skill-cards
Also available as an MCP server (npm install skill-tree-ai) for Cursor, VS Code, or Windsurf, and as a zip for Cowork.
github.com/robertnowell/skill-tree
Top comments (0)