Anthropoc published a study in February that classified 11 observable collaboration behaviors across 9,830 Claude conversations. I'd been using Claude Code daily for months and wanted to know how my sessions mapped against that distribution — not a score, but a breakdown: which behaviors showed up, which were absent, which I'd apparently never tried.
I built skill-tree to answer that. It reads your Claude Code session history, runs the same 11-behavior classification from the AI Fluency Index, then assigns one of seven archetype cards based on your actual distribution. More usefully: it picks the one behavior you haven't touched and surfaces it as a growth quest pinned to your next session via the SessionStart hook.
The behavior taxonomy comes from Dakan & Feller's 4D AI Fluency Framework — Description, Discernment, Delegation, and Diligence. Diligence doesn't show up in chat logs, so the classifier works across the other three axes. The 11 behaviors are the same ones Anthropic used for the population study, which means your output is directly comparable to that baseline.
End-to-end pipeline is 7 steps: find session files, extract user messages, remote classifier (Claude Haiku on Fly.io), archetype assignment, narrative synthesis, render, return a stable URL. Takes 30–60 seconds. The archetype cards are rendered as tarot cards using curated museum art — you can see a live example at skill-tree-ai.fly.dev/fixture/illuminator.
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 there's an MCP server (npm install skill-tree-ai) for Cursor, VS Code, and Windsurf.
Growth quest state persists across sessions. Claude Code uses ~/.skill-tree/; Cowork uses $CLAUDE_PLUGIN_ROOT/.user-state/ because its $HOME is ephemeral.
github.com/robertnowell/skill-tree
Top comments (0)