Agent Friendly Code
← back to leaderboard

karpathy/llm.cGitHub

https://github.com/karpathy/llm.c
Stars:
29.7k
Default branch:
master
Last scored:
7h ago
Overall

Strengths

  • CI configuration
  • Dependency manifest
  • Reproducible dev env
  • License file
  • README

Gaps

  • AGENTS.md / CLAUDE.md
  • .aider.conf.yml
  • CONTRIBUTING guide
  • Cursor rules (.cursor/rules)
  • GEMINI.md

Suggestions to improve for a specific model

  1. 1
    AGENTS.md / CLAUDE.md
    Add an AGENTS.md covering project goals, layout, setup commands, and conventions. Aim for 800+ chars of real guidance (not boilerplate).
    +13.2 pts
  2. 2
    Test suite
    Add a tests/ (or test/, __tests__/, spec/) directory with runnable tests. Document how to run them in AGENTS.md.
    +13.2 pts
  3. 3
    Linter / formatter config
    Configure a linter/formatter (ESLint+Prettier, Biome, Ruff, rustfmt+clippy, golangci-lint) and commit the config.
    +7.9 pts

Per-model scores

Claude Code
Loads CLAUDE.md at the start of every conversation per Anthropic's memory docs, so AGENTS.md / CLAUDE.md and a fast test loop carry the most weight.
47.4
Cursor
Per Cursor's Rules docs, reads `.cursor/rules/*.mdc` and AGENTS.md as the canonical repo-side input. Type config and a clean README still aid the codebase index but aren't the docs-cited signal.
41.0
Devin
Operates from a sandboxed Ubuntu VM and runs an 8-step machine setup (deps, secrets, language versions, lint/test commands) per Cognition's repo-setup docs. CI config files alone aren't what the docs ask for — a runnable dev environment is.
54.5
GPT-5 Codex
Reads AGENTS.md before doing any work per OpenAI's Codex docs — the strictest AGENTS.md adherent of any agent here. Hierarchical (per-directory) AGENTS.md and AGENTS.override.md are first-class.
49.3
Gemini CLI
Reads hierarchical `GEMINI.md` (global → workspace → component-level) at every prompt per Gemini CLI's docs. The long-context advantage favors repos that split context per directory rather than docs-heavy in general.
43.2
Aider
Auto-lints on every edit by default; runs the configured test command after edits when `--test-cmd` is set (per Aider's lint/test docs). A green linter and a declared test command translate directly into successful commits.
36.5
OpenHands
Runs in a sandboxed container and executes `.openhands/setup.sh` at session start per OpenHands' repo-customization docs. Root AGENTS.md is now the preferred always-on instruction surface (microagents are deprecated in favor of it).
50.0
Pi
Minimal terminal coding harness. Loads `AGENTS.md` (or `CLAUDE.md`) at startup — global, parent dirs, then cwd — per the Pi coding-agent README. Sandboxing is deferred to user-installed extensions.
43.7

Signal breakdown

AGENTS.md / CLAUDE.md· 0% pass
No agent instructions file found
.aider.conf.yml· 0% pass
No .aider.conf.yml at repo root
CI configuration· 100% pass
3 GitHub Actions workflow(s)
.github/workflows
CONTRIBUTING guide· 0% pass
No CONTRIBUTING file
Cursor rules (.cursor/rules)· 0% pass
No .cursor/rules/*.mdc or .cursorrules
Dependency manifest· 100% pass
Manifest present
requirements.txt
Reproducible dev env· 100% pass
2 env artifacts (Makefile, makefile)
Makefile
GEMINI.md· 0% pass
No GEMINI.md at repo root
License file· 100% pass
License present
LICENSE
Linter / formatter config· 0% pass
No linter/formatter config found
.openhands/setup.sh· 0% pass
No .openhands/setup.sh
Pre-commit / git hooks· 0% pass
No pre-commit / husky / lefthook found
README· 100% pass
README detailed (16466 chars)
README.md
Manageable size· 100% pass
98 files
Test suite· 0% pass
No test directory or test files found
Type configuration· 0% pass
No type config found

Embed this badge

Drop the snippet below into the top of this repo's README so visitors can see its agent-friendliness at a glance — clicks land back here.

Agent friendly score for github/karpathy/llm.c
[![Agent Friendly](https://agentfriendlycode.com/api/badge/github/karpathy/llm.c.svg)](https://agentfriendlycode.com/repo/136)