Cursor vs Codex: which do top repos set up for?
Codex averages higher: 63.5 for Cursor and 70.5 for Codex, across 402 repos. Cursor scores higher on 14, Codex on 388, and 0 tie.
Scores measure how well each repo is set up for an agent: instruction files, tests, CI and the other checks, weighted by what each agent's docs say it uses, not which agent writes better code.
Which files does each one read?
- Reads from the repo
- Cursor:
.cursor/rules/*.mdcAGENTS.mdCLAUDE.md.agents/skills/ - Codex:
AGENTS.mdAGENTS.override.md.agents/skills/ - Reads AGENTS.md
- Cursor: Always
- Codex: Always
- Reads CLAUDE.md
- Cursor: Yes
- Codex: No
- From its docs
- Cursor: Project rules are
.mdcfiles in.cursor/rules/, subfolders included; a plain.mdthere is ignored. AGENTS.md is read at the root and in subdirectories, combined, the more specific winning. CLAUDE.md is read the same way; the docs describe one at the root. A root.cursorrulesis legacy and will be deprecated. - Codex: Reads AGENTS.md before any work, one per directory from the repo root down to the working directory, up to 32 KiB combined; AGENTS.override.md replaces the AGENTS.md beside it.
- Repos with its own file
- Cursor: .cursor/rules or .cursorrules: 18 of 402 (4%)
- Codex: AGENTS.md: 207 of 402 (51%)
AGENTS.md and CLAUDE.md mean the file at the repo root; a repo with both counts under AGENTS.md, and empty files don't count. Every agent's files.
Which checks does each weigh more?
| Check | Cursor | Codex |
|---|---|---|
| Cursor rules (.cursor/rules) | 11% | 0% |
| CI configuration | 5% | 9% |
| Reproducible dev env | 6% | 9% |
| Type configuration | 11% | 9% |
| AGENTS.md / CLAUDE.md | 9% | 11% |
| Test suite | 8% | 10% |
Each figure is the check's share of the agent's total weight, from the docs linked below. A check few repos pass, such as an agent's own file, lowers that agent's scores wherever it carries weight. The methodology lists every weight.
Which repos are set up better for Cursor than Codex?
- TheOrcDev/8bitcn-ui78.4 vs 73.7 for Codex
- Shopify/cli89.5 vs 85.2 for Codex
- chenglou/pretext74.6 vs 70.6 for Codex
- mastra-ai/mastra91.6 vs 88.4 for Codex
- novuhq/novu90.5 vs 88.1 for Codex
Which repos are set up better for Codex than Cursor?
- prisma/orm73.4 vs 59.5 for Cursor
- hiyouga/LlamaFactory87.2 vs 74.4 for Cursor
- scikit-learn/scikit-learn81.8 vs 69.1 for Cursor
- github/spec-kit82.2 vs 69.8 for Cursor
- makeplane/plane67.7 vs 55.4 for Cursor
Which one leads in each language?
| Language | Repos | Cursor | Codex |
|---|---|---|---|
| TypeScript | 83 | 68.3 | 75.7 |
| Python | 56 | 66.9 | 74.1 |
| Rust | 45 | 67.8 | 74.0 |
| Go | 40 | 67.6 | 74.6 |
| Java | 28 | 58.4 | 65.5 |
| C++ | 21 | 54.9 | 62.4 |
| JavaScript | 15 | 69.4 | 77.1 |
What if I use both?
- Write the shared instructions once, in AGENTS.md at the repo root. What to put in it.
- Cursor: put rules in
.cursor/rules/*.mdc; a plain.mdthere is ignored. It reads AGENTS.md too. - Codex: reads AGENTS.md before any work, up to 32 KiB combined across directories.
AGENTS.md vs CLAUDE.md · Which files each agent reads · Make a repo agent-ready
From the docs: cursor.com/docs/rules (new tab) · cursor.com/skills (new tab) · cursor.com/context/rules (new tab) · cursor.com/customization/rules (new tab) · cursor.com/using (new tab) · learn.chatgpt.com/agents-md (new tab) · learn.chatgpt.com/build-skills (new tab)
Boards: best set up for Cursor · best set up for Codex
More comparisons: Claude Code vs Cursor · Claude Code vs Codex · Gemini CLI vs Claude Code · Aider vs Claude Code · OpenHands vs Claude Code · Pi vs Claude Code · Kimi Code CLI vs Claude Code · Devin vs Claude Code · Codex vs Gemini CLI