DEV Community

J Now
J Now

Posted on

What My Claude Code Habits Actually Look Like Against 9,830 Sessions

I've been using Claude Code daily for months and had no real signal on whether I was improving or just repeating the same interaction patterns faster. Anthropic published a study in February 2026 that measured 11 observable collaboration behaviors across 9,830 Claude conversations. I wanted to know what mine looked like against that baseline — specifically which behaviors I was hitting and which I never touched.

So I built skill-tree: it analyzes your Claude Code session history, classifies those same 11 behaviors from the AI Fluency Index, and returns a result as a hosted URL with a tarot-card-style archetype.

The taxonomy comes from Dakan & Feller's 4D AI Fluency Framework (Description, Discernment, Delegation, and Diligence — that fourth axis isn't visible in chat logs, so it's excluded). The classifier runs on Claude Haiku on Fly.io. End-to-end takes 30–60 seconds: find session files → extract user messages → remote classifier → archetype assignment → narrative synthesis → render → stable URL.

The output I actually cared about: a "growth quest" — one behavior from the 11 I hadn't used in recent sessions, surfaced as a concrete thing to try next time. That quest persists across sessions via a SessionStart hook, so Claude Code reminds me of it at the start of the next conversation.

To install in Claude Code:

claude plugin marketplace add robertnowell/ai-fluency-skill-cards
claude plugin install skill-tree-ai@ai-fluency-skill-cards
Enter fullscreen mode Exit fullscreen mode

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.

Live example of the archetype card output: https://skill-tree-ai.fly.dev/fixture/illuminator

Code: https://github.com/robertnowell/ai-fluency-skill-cards

Top comments (0)