README: how much a repo needs for AI coding agents
The README is the first file anyone opens, agents included. This check looks for one at the repo root and measures its length.
How many repos have a detailed README?
Of the 401 repos on the leaderboard, 384 pass in full (96%), 10 pass in part (2%) and 7 fail (2%).
Why do AI coding agents need a detailed README?
An agent starts every task knowing nothing about the project. A README that says what it does, how to install it and how the code is laid out saves it reading file by file. Cursor's profile gives the README its top weight.
What makes a good README for AI agents?
- Full pass: a README (
.md,.rst,.txtor no extension) with 1,000 or more characters. - 70%: 200 to 999 characters.
- 30%: under 200 characters.
- Fail: no README at the repo root.
How checks become a score: methodology.
How do I write a README AI agents can use?
Cover what the project does, how to install it, the commands people run most, and where the main code lives. That usually clears 1,000 characters.
Which AI coding agents weigh the README most?
- Cursor11% of its score
- Codex10% of its score
- Gemini CLI10% of its score
- Pi9% of its score
- Kimi Code CLI9% of its score
- Claude Code9% of its score
- Aider8% of its score
- Devin8% of its score
- OpenHands7% of its score
Pass rate by language
- Python56 of 56 (100%)
- Go40 of 40 (100%)
- JavaScript15 of 15 (100%)
- PHP12 of 12 (100%)
- Swift10 of 10 (100%)
- Dart7 of 7 (100%)
- Elixir7 of 7 (100%)
- Haskell7 of 7 (100%)
- Scala6 of 6 (100%)
- Clojure5 of 5 (100%)
- TypeScript80 of 82 (98%)
- Rust43 of 45 (96%)
- Kotlin11 of 12 (92%)
- C9 of 10 (90%)
- Ruby9 of 10 (90%)
- Java25 of 28 (89%)
- C#7 of 8 (88%)
- C++18 of 21 (86%)
- Zig5 of 6 (83%)
Top-scoring repos that pass
- Rust · 6.8k ★
- Ruby · 48.1k ★
- pingcap/tidb89.6Go · 40.7k ★
- gitlab-org/cli88.8Go · 1.1k ★
- apache/superset88.7Python · 75.4k ★
- axios/axios88.1JavaScript · 109.3k ★
- aio-libs/aiohttp88.1Python · 16.6k ★
- PostHog/posthog88.0Python · 40.7k ★
- C++ · 133.6k ★
- Python · 46.3k ★
- Rust · 33.3k ★
- TypeScript · 115 ★
Popular repos that fail it
- Java · 81.8k ★
- Java · 78.5k ★
- ocornut/imgui21.2C++ · 77k ★
- jekyll/jekyll50.3Ruby · 51.7k ★
- C# · 46.5k ★
- apache/cassandra36.7Java · 10.1k ★
- ocaml/ocaml40.0OCaml · 6.6k ★
Every check, grouped into what to do first: how to make your repo ready for AI coding agents.