I'd been shipping with Claude Code for months before I noticed the pattern: I was fast, but I was running the same three moves on repeat. Describe a problem, ask for code, iterate on the output. I never stopped to ask whether that covered the space of what experienced collaborators actually do.
Anthropics's AI Fluency Index (February 2026, 9,830 Claude conversations studied) classifies 11 observable behaviors across three axes — Description, Discernment, and Delegation. Delegation is things like explicitly assigning Claude a role, asking it to critique its own output, or handing off a multi-step plan and checking back at milestones. I had assumed I did these. I didn't.
skill-tree reads your Claude Code session history, classifies those same 11 behaviors against your actual messages, and renders the results as a tarot-card archetype with a skill radar. Seven archetypes total — Illuminator, Navigator, Alchemist, and four others. Each is rendered with curated museum art; the live Illuminator card is at skill-tree-ai.fly.dev/fixture/illuminator if you want to see what the output looks like.
The part I find most useful isn't the archetype label. It's the growth quest: one behavior from the 11 that your sessions haven't surfaced, chosen specifically because it's absent from your history, and persisted via a SessionStart hook so it shows up at the top of your next session.
The 7-step orchestration (find session files → extract user messages → remote classifier → archetype assignment → narrative synthesis → render → return URL) runs in 30–60 seconds. Classifier runs on Claude Haiku on Fly.io; the rendered card gets a stable URL stored on a Fly.io volume.
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, and Windsurf, and as a zip for Cowork.
github.com/robertnowell/skill-tree
Top comments (0)