openbook
a personality card built from your Claude Code + Codex prompt history. what your AI assistants have inferred about you, with receipts.
Problem
Claude Code and Codex sit alongside engineers every day. Across thousands of sessions they build up an implicit model of who you are — what you know, what you don't, what you ask about, how you ask, what you avoid. That model shapes their replies. It's invisible to the user.
I wanted to see what mine looked like.
Approach
Openbook parses your local Claude Code + Codex prompt history (it never leaves your machine), runs analysis over it, and produces a personality card: interests, technical skill level, working style, recurring themes — each backed by the actual prompts as receipts so you can audit why it inferred what it inferred.
- Local-only — your prompt history is your data; openbook reads it from disk and analyses in-process.
- Receipts — every claim links back to the prompts that justified it. No vibes, no opaque scoring.
- Open source — see exactly what's being extracted and how.
Today
Open source, usable. Early — the analysis surface is small and the UI is a static HTML card. Want to see what an AI assistant has been quietly modelling about you? Clone and run.