Methodology
How the leaderboard is built, and what the score does and does not mean.
How we score
Each repo is shallow-cloned, scanned by @agentlinthq/cli against the same rubric your repo would be scored by, and ranked. We do not weight, reweight, or curate the rankings. The CLI is open source; the weights are part of the public scoring API and changing them is a breaking change. Source →
Cadence
The scan runs weekly on Mondays at 06:00 UTC against the top 100 most-starred public GitHub repos. The aggregated JSON is committed to the public CLI repo; this page reads it directly. Nothing is cached except the standard Next.js edge cache (one day).
Anti-gaming clause
A high agentlint score means the repo's signals look agent-ready. It does not mean the code is good, the project is healthy, or the maintainers are responsive. The score is what the score is.