{"name":"Repo Agent Kit Agent Readiness Benchmark","snapshot":{"date":"2026-08-31","sampleSize":10,"averageScore":62,"rootAgentsCount":4,"agentReadyCount":2},"methodology":{"scope":"Root files plus direct children of .github and .cursor at the pinned commit.","weights":{"Root AGENTS.md or tool-specific guidance":30,"AGENTS.md instruction quality":20,"Root README":10,"Build manifest":10,"Dependency lockfile":8,"Visible root test surface":8,"Recognized CI surface":7,"Contribution, security, or ownership guidance":7},"limitation":"This static audit measures discoverable repository evidence. It does not execute code, inspect nested package quality, or predict task success."},"repositories":[{"fullName":"anthropics/claude-code","branch":"main","commit":"f275fa282e76c5e5456912268f2c367a7f4f4797","score":24,"grade":"Not agent-ready","instructionScore":null,"evidence":{"guidance":null,"readme":"README.md","manifest":null,"lockfile":null,"tests":null,"ci":".github/workflows","maintenance":"SECURITY.md"}},{"fullName":"django/django","branch":"main","commit":"189136c2a3e166c59a49cca2444aa6a1d77aa136","score":54,"grade":"Needs guidance","instructionScore":null,"evidence":{"guidance":".github/copilot-instructions.md","readme":"README.rst","manifest":"pyproject.toml","lockfile":null,"tests":"tests","ci":".github/workflows","maintenance":".github/SECURITY.md"}},{"fullName":"facebook/react","branch":"main","commit":"065bc84eb50914fa1bc60983eac0b0b717186d74","score":54,"grade":"Needs guidance","instructionScore":null,"evidence":{"guidance":"CLAUDE.md","readme":"README.md","manifest":"package.json","lockfile":"yarn.lock","tests":null,"ci":".github/workflows","maintenance":"CONTRIBUTING.md"}},{"fullName":"golang/go","branch":"master","commit":"f534300f98306366cb221806394b223d3b30eb51","score":25,"grade":"Not agent-ready","instructionScore":null,"evidence":{"guidance":null,"readme":"README.md","manifest":null,"lockfile":null,"tests":"test","ci":null,"maintenance":"CONTRIBUTING.md"}},{"fullName":"microsoft/vscode","branch":"main","commit":"bcec3677a1c13f4e33d8ad49a8d9fa4fd65383d9","score":85,"grade":"Good foundation","instructionScore":26,"detail":{"slug":"microsoft-vscode","wordCount":33,"lineCount":6,"passed":["Repository purpose","Architecture map"],"notDetected":["Stack and runtime","Runnable commands","Validation loop","Change boundaries","Safety guardrails","Definition of done"],"focusedLength":false,"context":"The root file is a short pointer to .github/copilot-instructions.md. The deterministic file-quality score evaluates AGENTS.md itself and deliberately does not merge delegated files, so read this result as a delegation pattern rather than a complete instruction audit."},"evidence":{"guidance":"AGENTS.md","readme":"README.md","manifest":"package.json","lockfile":"package-lock.json","tests":"test","ci":".github/workflows","maintenance":"CONTRIBUTING.md"}},{"fullName":"openai/codex","branch":"main","commit":"13bc770eaf0ad8548776bde59c3d6e5316406279","score":83,"grade":"Good foundation","instructionScore":54,"detail":{"slug":"openai-codex","wordCount":3141,"lineCount":323,"passed":["Repository purpose","Stack and runtime","Runnable commands","Architecture map"],"notDetected":["Validation loop","Change boundaries","Safety guardrails","Definition of done"],"focusedLength":false,"context":"The root file contains extensive, repository-specific Rust and Codex guidance. It exceeds this benchmark’s focused-root guideline, and the phrase-based checks do not treat every domain-specific rule as a match for a generic category."},"evidence":{"guidance":"AGENTS.md","readme":"README.md","manifest":"package.json","lockfile":"pnpm-lock.yaml","tests":null,"ci":".github/workflows","maintenance":"SECURITY.md"}},{"fullName":"rust-lang/rust","branch":"main","commit":"0dfb098f3aeecbe38c2566ca090193280e7349e7","score":94,"grade":"Agent-ready","instructionScore":70,"detail":{"slug":"rust-lang-rust","wordCount":1747,"lineCount":244,"passed":["Repository purpose","Stack and runtime","Runnable commands","Architecture map","Validation loop"],"notDetected":["Change boundaries","Safety guardrails","Definition of done"],"focusedLength":false,"context":"The root file is a strict operational policy with repository routing and validation guidance. Its length is above the focused-root guideline, while several domain-specific gates do not match the benchmark’s generic phrases."},"evidence":{"guidance":"AGENTS.md","readme":"README.md","manifest":"Cargo.toml","lockfile":"Cargo.lock","tests":"tests","ci":".github/workflows","maintenance":"CONTRIBUTING.md"}},{"fullName":"tiangolo/fastapi","branch":"master","commit":"49033471594ea5d99a80abdf1043231b7791ee49","score":43,"grade":"Not agent-ready","instructionScore":null,"evidence":{"guidance":null,"readme":"README.md","manifest":"pyproject.toml","lockfile":"uv.lock","tests":"tests","ci":".github/workflows","maintenance":null}},{"fullName":"vercel/next.js","branch":"canary","commit":"d434afa837b995d2db6f101e3b705461a250c655","score":96,"grade":"Agent-ready","instructionScore":82,"detail":{"slug":"vercel-nextjs","wordCount":4334,"lineCount":561,"passed":["Repository purpose","Stack and runtime","Runnable commands","Architecture map","Validation loop","Safety guardrails"],"notDetected":["Change boundaries","Definition of done"],"focusedLength":false,"context":"The root file is comprehensive and covers many workspace-specific workflows. It scores strongly on the deterministic categories but is much longer than the benchmark’s focused-root guideline."},"evidence":{"guidance":"AGENTS.md","readme":"readme.md","manifest":"package.json","lockfile":"pnpm-lock.yaml","tests":"jest.config.js","ci":".github/workflows","maintenance":"contributing.md"}},{"fullName":"vitejs/vite","branch":"main","commit":"238ad811c7fb9e4730cbd317d0657867ed3447b3","score":62,"grade":"Needs guidance","instructionScore":null,"evidence":{"guidance":".github/copilot-instructions.md","readme":"README.md","manifest":"package.json","lockfile":"pnpm-lock.yaml","tests":"vitest.config.e2e.ts","ci":".github/workflows","maintenance":"CONTRIBUTING.md"}}]}