Get your score
Pick how you want to be assessed. All four methods feed into the same scoring function and produce a comparable score on the leaderboard.
16 multiple-choice questions across all 8 dimensions. Client-side scoring, no signup.
Scan a public repo for context-management signals. Read-only, no source code stored.
Run npx context-index in any local repo. Local-first, share only the result hash.
Drop CLAUDE.md, .cursorrules, AGENTS.md, README. Analyzed entirely client-side.
GitHub scans store only metadata signals (does CLAUDE.md exist? line count? last modified?), never source code. File uploads and CLI scans run entirely client-side or locally. The composite score and rank are public; per-dimension breakdowns require an email so we can email you historical comparisons.