For months I used Claude Code daily and assumed I was improving. Then I read Anthropic's February 2026 study — 9,830 conversations, 11 observable collaboration behaviors measured across a real user population — and realized I had no idea which of those 11 behaviors I actually used, let alone which ones I never touched.
skill-tree reads your Claude Code or Cowork session history, classifies the same 11 behaviors against that baseline, and assigns you one of seven archetype cards.
The archetypes are rendered as tarot cards — Illuminator, Navigator, Alchemist, and four others — with curated museum art pulled for each. They're not aesthetic decoration. Each card maps to a specific behavioral profile derived from Dakan & Feller's 4D AI Fluency Framework (Description, Discernment, Delegation — Diligence is the fourth axis but doesn't show up in chat logs, so it's excluded from scoring). The radar on each card shows exactly where you sit across the three measurable axes.
Live example, rendered from fixture data: 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 tried and surfaces it as a concrete prompt pattern for your next session. It persists across sessions via a SessionStart hook — Claude Code writes to ~/.skill-tree/, Cowork writes to $CLAUDE_PLUGIN_ROOT/.user-state/ because its $HOME is ephemeral. So the quest follows you session to session until you close it out.
Install in Claude Code:
claude plugin marketplace add robertnowell/ai-fluency-skill-cards && claude plugin install skill-tree-ai@ai-fluency-skill-cards
Or use the MCP server (npm install skill-tree-ai) if you're on Cursor, VS Code, or Windsurf.
The classifier runs on Fly.io (Claude Haiku), full analysis takes 30–60 seconds, and you get back a stable URL to your rendered card.
Top comments (0)