GEMINI.md: the instructions file Gemini CLI reads
GEMINI.md is Gemini CLI's AGENTS.md: project goals, layout, setup commands and conventions, in markdown.
How many repos have a GEMINI.md?
Of the 401 repos on the leaderboard, 6 pass in full (1%), 2 pass in part (0.5%) and 393 fail (98%).
Why does Gemini CLI need a GEMINI.md?
Gemini CLI's docs describe GEMINI.md as layered, from a global file down to one per component, and sent with every prompt. Of the agents we score, Gemini CLI reads it (OpenHands only when it runs a Gemini model), so this check only affects the Gemini CLI score.
What makes a good GEMINI.md?
- Full pass: GEMINI.md at the repo root with 800 or more characters, counting an AGENTS.md, CLAUDE.md or AGENT.md it imports with a line like
@AGENTS.md. - 80%: 200 to 799 characters.
- 50%: under 200 characters.
- 20%: the file exists but is empty.
- Fail: no GEMINI.md at the repo root.
How checks become a score: methodology.
How do I add a GEMINI.md to my repo?
Add a GEMINI.md at the repo root with what you would put in AGENTS.md, or a single @AGENTS.md line to reuse that file. We check only the root file, though Gemini CLI also reads them in subdirectories.
Which AI agents read GEMINI.md?
- Gemini CLI11% of its score
Pass rate by language
- Ruby1 of 10 (10%)
- Rust2 of 45 (4%)
- Java1 of 28 (4%)
- Python1 of 56 (2%)
- TypeScript1 of 82 (1%)
- Go0 of 40 (0%)
- C++0 of 21 (0%)
- JavaScript0 of 15 (0%)
- Kotlin0 of 12 (0%)
- PHP0 of 12 (0%)
- C0 of 10 (0%)
- Swift0 of 10 (0%)
- C#0 of 8 (0%)
- Dart0 of 7 (0%)
- Elixir0 of 7 (0%)
- Haskell0 of 7 (0%)
- Scala0 of 6 (0%)
- Zig0 of 6 (0%)
- Clojure0 of 5 (0%)
GEMINI.md vs AGENTS.md: which to use, and how to keep one source.
Top-scoring repos that pass
- Ruby · 48.1k ★
- apache/superset88.7Python · 75.4k ★
- Rust · 94k ★
- TypeScript · 108.8k ★
- Rust · 128.5k ★
- apache/seatunnel64.4Java · 9.8k ★
Popular repos that fail it
- TypeScript · 394.1k ★
- react/react64.1JavaScript · 252.3k ★
- TypeScript · 225.1k ★
- n8n-io/n8n83.6TypeScript · 209.7k ★
- TypeScript · 199.4k ★
- microsoft/vscode72.0TypeScript · 195.1k ★
- Python · 187.9k ★
- ollama/ollama76.7Go · 186.2k ★
Every check, grouped into what to do first: how to make your repo ready for AI coding agents.