llms.txt
Compare with other sites' llms.txt files →# Agent Friendly Code
> Scores how ready a repo is for AI coding agents, one score per agent. Ranked public repos, a live score by URL, an agent skill and a GitHub Action.
Every score comes from 17 file checks, listed below; no agent is run. Each AI coding agent weighs the checks by what its own docs say it reads, so the same repository can score differently for different agents.
## Agents scored for
- [Claude Code](https://agentfriendlycode.com/model/claude-code): Loads CLAUDE.md (or `.claude/CLAUDE.md`) at the start of every session per Anthropic's memory docs, and AGENTS.md only when there is no CLAUDE.md, so the instructions file and a fast test loop carry the most weight. Its skills live in `.claude/skills/`, not `.agents/skills/`.
- [Cursor](https://agentfriendlycode.com/model/cursor): Per Cursor's Rules docs, reads `.cursor/rules/*.mdc` (subfolders included) and AGENTS.md as its repo instructions, and loads skills from `.agents/skills/`. Type config and a clear README help its codebase search but aren't the docs-cited signal.
- [Devin](https://agentfriendlycode.com/model/devin): Works in a Linux virtual machine booted from a snapshot that already has the repo cloned and its tools, dependencies, environment variables and secrets set up, then installs, lints, builds and tests like any developer, per Cognition's environment docs; a committed `.devin/blueprint.yaml` defines that environment. CI config files alone aren't what the docs ask for — a runnable dev environment is. Pulls Cursor rule files into its Knowledge and discovers skills in `.agents/skills/`.
- [Codex](https://agentfriendlycode.com/model/gpt-5-codex): Reads AGENTS.md before doing any work per OpenAI's Codex docs: one per directory from the repo root down to the working directory, where AGENTS.override.md replaces AGENTS.md, up to 32 KiB combined. Skills load from `.agents/skills/`.
- [Kimi Code CLI](https://agentfriendlycode.com/model/kimi-cli): Reads AGENTS.md as its native instruction surface (root or `.kimi-code/AGENTS.md`) per Moonshot's Kimi Code docs, and `/init` generates one. Its docs name no CLAUDE.md fallback, but the check passes on CLAUDE.md too. Skills load from `.agents/skills/`. Runs shell commands step-by-step under approval modes rather than in a sandbox VM, and dispatches `explore` sub-agents with isolated contexts to map a codebase, so a large tree costs it less than a single-context agent.
- [Gemini CLI](https://agentfriendlycode.com/model/gemini-cli): Loads `GEMINI.md` from the global file and from the workspace and its parent directories and sends it with every prompt, per Gemini CLI's docs; one in a subdirectory loads when a tool first touches that directory. AGENTS.md is read only when listed in the `context.fileName` setting. Skills load from `.agents/skills/`.
- [Aider](https://agentfriendlycode.com/model/aider): Auto-lints on every edit by default, and runs the test command after edits once both `test-cmd` and `auto-test` are set (per Aider's lint/test docs). It doesn't natively read AGENTS.md or CONVENTIONS.md — those load only when wired via `.aider.conf.yml`'s `read:` — so the config file, a green linter, and a declared test command are what translate into successful commits.
- [OpenHands](https://agentfriendlycode.com/model/openhands): Executes `.openhands/setup.sh` whenever it starts work on a repo, per OpenHands' repo-customization docs; its backend runs on the user's machine by default, or in Docker, a VM or OpenHands Cloud. A root AGENTS.md goes into its system prompt in full; skills live in `.agents/skills/`, with the legacy `.openhands/skills/` and `.openhands/microagents/` still loaded.
- [Pi](https://agentfriendlycode.com/model/pi): Minimal terminal coding harness. Loads `AGENTS.md` (or `CLAUDE.md`) from its agent directory, the working directory and its parent directories, per Pi's configuration docs, and skills from `.agents/skills/`. It has no built-in sandbox; isolation is left to a container the user runs it in.
## Checks
- [AGENTS.md / CLAUDE.md](https://agentfriendlycode.com/check/agents-md): What an AGENTS.md or CLAUDE.md file is, which AI coding agents read it, how we score it, and how many popular open source repos already have one.
- [Cursor rules (.cursor/rules)](https://agentfriendlycode.com/check/cursor-rules): What Cursor rule files are, the difference between .cursor/rules/*.mdc and the older .cursorrules file, how we score them, and which repos ship them.
- [GEMINI.md](https://agentfriendlycode.com/check/gemini-md): What GEMINI.md is, how Gemini CLI uses it, how we score it, and how many popular open source repos have one.
- [.openhands/setup.sh](https://agentfriendlycode.com/check/openhands-setup): What .openhands/setup.sh does, when OpenHands runs it, how we score it, and how many popular open source repos have one.
- [.aider.conf.yml](https://agentfriendlycode.com/check/aider-conf): What .aider.conf.yml is, which settings matter for an agent-ready repo, how we score it, and how many popular open source repos have one.
- [Skills (.agents/skills)](https://agentfriendlycode.com/check/agent-skills): What agent skills in .agents/skills are, which AI coding agents load them, how we score them, and how many popular open source repos ship them.
- [README](https://agentfriendlycode.com/check/readme): Why a detailed README helps AI coding agents, how we score README length, and how popular open source repos compare.
- [Test suite](https://agentfriendlycode.com/check/tests): Why AI coding agents depend on a test suite, how we detect tests in a repo, and how many popular open source repos pass.
- [CI configuration](https://agentfriendlycode.com/check/ci): Why a CI pipeline helps AI coding agents, which CI systems we detect, and how many popular open source repos have one.
- [Linter / formatter config](https://agentfriendlycode.com/check/linter): Why a committed linter or formatter config helps AI coding agents, which tools we detect, and how many popular open source repos have one.
- [Dependency manifest](https://agentfriendlycode.com/check/deps-manifest): Why AI coding agents need a dependency manifest, which manifest files we detect, and how many popular open source repos have one.
- [Reproducible dev env](https://agentfriendlycode.com/check/dev-env): Why AI coding agents need a reproducible dev environment, which setup files we detect, and how many popular open source repos have them.
- [Type configuration](https://agentfriendlycode.com/check/type-config): Why static types help AI coding agents, how we detect type checking in a repo, and how many popular open source repos have it.
- [License file](https://agentfriendlycode.com/check/license): Why a license file matters when AI coding agents work in a repo, which files we detect, and how many popular open source repos have one.
- [CONTRIBUTING guide](https://agentfriendlycode.com/check/contributing): Why a CONTRIBUTING guide helps AI coding agents, where we look for one, and how many popular open source repos have it.
- [Pre-commit / git hooks](https://agentfriendlycode.com/check/pre-commit): Why git hooks help AI coding agents, which hook frameworks we detect, and how many popular open source repos use one.
- [Manageable size](https://agentfriendlycode.com/check/size): Why very large repos are harder for AI coding agents, how we count files, and how popular open source repos are spread across the size bands.
## Key pages
- [Leaderboard](https://agentfriendlycode.com/): Public repos ranked overall and per agent, across GitHub, GitLab, and Bitbucket
- [About](https://agentfriendlycode.com/about): Who built this, why, and the project's independence statement
- [Action](https://agentfriendlycode.com/action): GitHub Action that comments on each pull request with the score change — install snippet, FAQ
- [Changelog](https://agentfriendlycode.com/changelog): What shipped per release
- [Guide: make a repo ready for AI coding agents](https://agentfriendlycode.com/guide): The checks grouped into what to do, with leaderboard pass rates and links to every check, agent file and tool
- [State of AGENTS.md adoption](https://agentfriendlycode.com/report/agents-md): Share of leaderboard repos with AGENTS.md or CLAUDE.md, from snapshots on most days since 24 Apr 2026; today's split by language and other agent files; CSV (https://agentfriendlycode.com/report/agents-md.csv)
- [Checks](https://agentfriendlycode.com/check): Agent-ready codebase checklist — every check, with what it looks for, how to pass it and how many leaderboard repos do
- [Which files each agent reads](https://agentfriendlycode.com/agents): AGENTS.md, CLAUDE.md, GEMINI.md, .cursor/rules and the rest, per agent, from each agent's docs, with how many leaderboard repos have each file
- [AGENTS.md vs CLAUDE.md](https://agentfriendlycode.com/check/agents-md/vs-claude-md): Which one to use, which agents read each, and what Claude Code does when a repo has both
- [Cursor rules vs AGENTS.md](https://agentfriendlycode.com/check/cursor-rules/vs-agents-md): Cursor reads both. AGENTS.md is plain markdown most agents read; .cursor/rules/*.mdc adds globs and descriptions so a rule loads only when it applies.
- [GEMINI.md vs AGENTS.md](https://agentfriendlycode.com/check/gemini-md/vs-agents-md): GEMINI.md is Gemini CLI's context file, sent with every prompt. Reuse AGENTS.md: list it in context.fileName, or in a one-line GEMINI.md: @AGENTS.md.
- [AGENTS.md examples](https://agentfriendlycode.com/check/agents-md/examples): Every AGENTS.md and CLAUDE.md on the leaderboard, linked to the file, with length, stars and score; by language: [TypeScript](https://agentfriendlycode.com/check/agents-md/examples/typescript), [Python](https://agentfriendlycode.com/check/agents-md/examples/python), [Rust](https://agentfriendlycode.com/check/agents-md/examples/rust), [Go](https://agentfriendlycode.com/check/agents-md/examples/go), [Java](https://agentfriendlycode.com/check/agents-md/examples/java), [C++](https://agentfriendlycode.com/check/agents-md/examples/cpp), [JavaScript](https://agentfriendlycode.com/check/agents-md/examples/javascript), [Kotlin](https://agentfriendlycode.com/check/agents-md/examples/kotlin), [PHP](https://agentfriendlycode.com/check/agents-md/examples/php), [Ruby](https://agentfriendlycode.com/check/agents-md/examples/ruby); by framework, with what to put in the file: [Next.js](https://agentfriendlycode.com/check/agents-md/examples/nextjs), [FastAPI](https://agentfriendlycode.com/check/agents-md/examples/fastapi), [Django](https://agentfriendlycode.com/check/agents-md/examples/django), [Rails](https://agentfriendlycode.com/check/agents-md/examples/rails)
- [Languages](https://agentfriendlycode.com/language): The leaderboard split by programming language, with each language's most-missed checks
- [Other languages](https://agentfriendlycode.com/language/other): Repos in languages too small for their own hub, grouped by language
- [Topics](https://agentfriendlycode.com/topics): The leaderboard split by repo topic, one ranked page per topic
- [Methodology](https://agentfriendlycode.com/methodology): How scores are computed; signals, weights, and limitations
- [Package lookup](https://agentfriendlycode.com/package): Resolve npm / PyPI / Cargo packages to their source-repo agent-friendliness score
- [Live Score](https://agentfriendlycode.com/score): Score any public GitHub repository on demand, including repos the leaderboard has not indexed
- [Roadmap](https://agentfriendlycode.com/roadmap): Upcoming versions
- [Skill](https://agentfriendlycode.com/skill): Agent skill that scores the repo you are in — install snippet, the agent your repo is best set up for, score → model class, optional SessionStart hooks for Claude Code / Codex
- [Sitemap](https://agentfriendlycode.com/sitemap.xml): Every indexed URL
## Public API
- [`GET /api/badge/{host}/{owner}/{name}.svg`](https://agentfriendlycode.com/api/badge/github/vercel/next.js.svg): Embeddable SVG badge (`?model=<id>` for one agent's score)
- [`GET /api/package/{registry}/{name}`](https://agentfriendlycode.com/api/package/npm/next): Resolve npm / PyPI / Cargo package → source-repo score
- [`GET /api/repos`](https://agentfriendlycode.com/api/repos): JSON array of every ranked repo: the repo row plus its overall score
- [`GET /api/repo/{id}`](https://agentfriendlycode.com/api/repo/1): Per-repo detail by internal id — `{ repo, signals, modelScores }`
- [`GET /api/score?host=&repo=owner/name`](https://agentfriendlycode.com/api/score?host=github&repo=vercel/next.js): Look up an indexed repo by host + owner/name — public read for external integrators (the PR-diff Action and the agent skill both vendor the scorer locally and don't call this at runtime)
## Tools
- [Agent Friendly Action](https://github.com/hsnice16/agent-friendly-action): GitHub Action that posts a per-PR score-delta comment — runs entirely inside the maintainer's CI, opt-in via `AGENTS_BADGE_TOKEN`, listed on the GitHub Marketplace
- [Agent Friendly Skill](https://github.com/hsnice16/agent-friendly-skill): Portable agent skill (Claude Code, Codex, Cursor, Cline, Copilot, …) installable via `npx skills add hsnice16/agent-friendly-skill#v0` — scores the user's current repo locally, no service dependency. Listed on [skills.sh](https://www.skills.sh/hsnice16/agent-friendly-skill/agent-friendly)
## Top scored packages
Each links to the score page of the package's source repo.
### npm
- [npm/axios](https://agentfriendlycode.com/repo/github/axios/axios): source repo axios/axios, 88.1 / 100
- [npm/next](https://agentfriendlycode.com/repo/github/vercel/next.js): source repo vercel/next.js, 86.9 / 100
- [npm/eslint](https://agentfriendlycode.com/repo/github/eslint/eslint): source repo eslint/eslint, 71.1 / 100
- [npm/prettier](https://agentfriendlycode.com/repo/github/prettier/prettier): source repo prettier/prettier, 67.9 / 100
- [npm/react](https://agentfriendlycode.com/repo/github/react/react): source repo react/react, 64.1 / 100
### pypi
- [pypi/requests](https://agentfriendlycode.com/repo/github/psf/requests): source repo psf/requests, 81.3 / 100
- [pypi/fastapi](https://agentfriendlycode.com/repo/github/fastapi/fastapi): source repo fastapi/fastapi, 64.4 / 100
### cargo
- [cargo/ripgrep](https://agentfriendlycode.com/repo/github/BurntSushi/ripgrep): source repo BurntSushi/ripgrep, 67.4 / 100
- [cargo/axum](https://agentfriendlycode.com/repo/github/tokio-rs/axum): source repo tokio-rs/axum, 55.6 / 100
- [cargo/tokio](https://agentfriendlycode.com/repo/github/tokio-rs/tokio): source repo tokio-rs/tokio, 54.4 / 100
- [cargo/serde](https://agentfriendlycode.com/repo/github/serde-rs/serde): source repo serde-rs/serde, 44.4 / 100
## Source
- [Repository](https://github.com/hsnice16/agent-friendly-code): MIT-licensed Next.js app
- [License](https://opensource.org/licenses/MIT): MIT