Which public repos are easiest for GPT-5 Codex to work in?
Repos from GitHub, GitLab, and Bitbucket, ranked by their GPT-5 Codex score rather than the average across every agent. gitlab-org/cli leads with 97.7.
The methodology lists the weight GPT-5 Codex gives each check and the documentation it comes from.
GPT-5 Codex — Reads AGENTS.md before doing any work per OpenAI's Codex docs. Hierarchical (per-directory) AGENTS.md and AGENTS.override.md are first-class.
| Rank | Repo | Stars | Score | External link |
|---|---|---|---|---|
| 353 | symfony/symfonyGitHub | 31.2k ★ | open | |
| 354 | mattermost/mattermostGitHub | 39.3k ★ | open | |
| 355 | opencv/opencvGitHub | 91.2k ★ | open | |
| 356 | ocaml/ocamlGitHub | 6.6k ★ | open | |
| 357 | inkscape/inkscapeGitLab | 4k ★ | open | |
| 358 | AvaloniaUI/AvaloniaGitHub | 31.7k ★ | open | |
| 359 | microsoft/semantic-kernelGitHub | 28.7k ★ | open | |
| 360 | karpathy/llm.cGitHub | 31.1k ★ | open | |
| 361 | stripe/aiGitHub | 1.9k ★ | open | |
| 362 | nim-lang/NimGitHub | 18.3k ★ | open | |
| 363 | apache/cassandraGitHub | 10.1k ★ | open | |
| 364 | karpathy/nanochatGitHub | 58.6k ★ | open | |
| 365 | ziglang/zigGitHub | 43.3k ★ | open | |
| 366 | farzaa/clickyGitHub | 7.7k ★ | open | |
| 367 | golang/goGitHub | 139.2k ★ | open | |
| 368 | multicoreware/x265_gitBitbucket | — | open | |
| 369 | microsoft/AIGitHub | 2.9k ★ | open | |
| 370 | ocornut/imguiGitHub | 76.6k ★ | open | |
| 371 | karpathy/nanoGPTGitHub | 63.7k ★ | open |
Tracking 371 repos · last updated