Anthropics February 2026 study classified 11 observable collaboration behaviors across 9,830 Claude conversations. I wanted to run the same classification on my own sessions — not to benchmark myself against other users, but to see which behaviors I was avoiding entirely.
The pattern I kept finding: heavy on Description (context-setting, explaining requirements), almost nothing on Delegation (trusting Claude to make structural decisions autonomously). I had optimized for speed at the exact same habits.
skill-tree pulls your Claude Code or Cowork session history, runs the same 11-behavior taxonomy from Dakan & Feller's 4D AI Fluency Framework, and returns a stable URL with your results. The classifier runs on Fly.io (~30–60 seconds end-to-end), and the visualization is stored on a volume so the URL persists.
The output includes a skill radar across three visible axes (Description, Discernment, Delegation — Diligence is the fourth axis but isn't recoverable from chat logs), one of seven archetype cards rendered as tarot cards with curated museum art, and a growth quest: a specific behavior from the 11 you haven't touched recently, surfaced as a prompt for your next session. The quest persists across sessions via a SessionStart hook.
The archetypes — Illuminator, Navigator, Alchemist, and four others — aren't just labels. Each maps to a specific behavioral signature across the three axes. You can see what the Illuminator card looks like rendered 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 as an MCP server (npm install skill-tree-ai) for Cursor, VS Code, or Windsurf.
github.com/robertnowell/skill-tree
Top comments (0)