After months of daily Claude Code sessions I noticed I was filing the same patterns over and over — tight prompts, quick iterations, same mental loop. Fast, but not necessarily growing. Anthropic published a study in February (9,830 conversations, 11 observable collaboration behaviors) that gave me a concrete vocabulary for what "better" could look like. I wanted my own sessions scored against that baseline.
skill-tree plugs into Claude Code and classifies your session history against those same 11 behaviors, pulled from Dakan & Feller's 4D AI Fluency Framework (Description, Discernment, Delegation — Diligence is the fourth axis, but it doesn't show up in chat logs). The output is a skill radar showing which behaviors you use and which you skip entirely.
The archetype assignment is where it gets interesting. Seven archetypes — Illuminator, Alchemist, Navigator, and four others — rendered as tarot cards using curated museum art. Each card is a stable URL you can share; the live fixture is at skill-tree-ai.fly.dev/fixture/illuminator. The whole pipeline (find session files → extract user messages → remote classifier → archetype assignment → narrative → render → return URL) runs in 30–60 seconds on Fly.io using Claude Haiku.
The growth quest is the part I built it for: one behavior you've never triggered, surfaced automatically and persisted via a SessionStart hook into ~/.skill-tree/ so it follows you across sessions.
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.
github.com/robertnowell/skill-tree
Top comments (0)