Spent months in Claude Code before I noticed I was stuck in a loop: describe a problem, get code, paste it in, repeat. Fast, but not necessarily better. Anthropic published a study in February (9,830 conversations, 11 observable collaboration behaviors) and I wanted to see my own sessions mapped against that baseline — not aggregated impressions, but actual classified behavior counts.
So I built skill-tree: it reads your Claude Code or Cowork session history, runs it through a remote classifier that tags each of the 11 behaviors from the AI Fluency Index, and returns a stable hosted URL with a skill radar and one of seven archetype cards.
The archetypes are rendered as tarot cards using curated museum art — Illuminator, Navigator, Alchemist, and four others. They're not aesthetic decoration. Each card maps to a behavior cluster across three axes from Dakan & Feller's 4D AI Fluency Framework: Description (how you frame problems), Discernment (how you evaluate outputs), Delegation (how you assign work). The fourth axis, Diligence, isn't measurable from chat logs so it's excluded.
You can see a live example at skill-tree-ai.fly.dev/fixture/illuminator.
The part I use most: the growth quest. After classification, the tool picks one behavior you haven't touched and surfaces it at the start of your next session via a SessionStart hook. It persists in ~/.skill-tree/ between sessions, so it doesn't reset on you.
End-to-end takes 30–60 seconds: find session files → extract user messages → remote classifier on Fly.io (Claude Haiku) → archetype assignment → narrative synthesis → render → return URL.
Install in Claude Code:
claude plugin marketplace add robertnowell/ai-fluency-skill-cards
claude plugin install skill-tree-ai@ai-fluency-skill-cards
Also works in Cowork via skill-tree-ai.zip, and as an MCP server (npm install skill-tree-ai) for Cursor, VS Code, and Windsurf.
Top comments (0)