multica-ai/andrej-karpathy-skills
A single CLAUDE.md distilling Andrej Karpathy's public observations on LLM coding pitfalls into instruction-file rules. An instruction set, not a skill framework. Agentiquette scores it 61/100 (Experimental-tier), methodology v1.0, scored 2026-07-07.
Key facts
| Category | Instruction set |
|---|---|
| License | No license file at fetch time |
| Maintainer | multica-ai |
| Compatible agents | claude-code |
| File patterns | CLAUDE.md |
| Triggering | auto-trigger |
| GitHub metrics | 188,780 stars · 19,389 forks · 124 open issues (fetched 2026-07-07) |
| Last commit | 2026-04-20 |
| Created | 2026-01-27 |
Editorial analysis
The content is better than the name-borrowing suggests: the rules target real, recurring failure modes (over-eager abstraction, silent assumption-making, unverified success claims). But it is one file with no update mechanism, no tests, and adoption metrics inflated by association. Scored as what it is: a good instruction set, experimental tier by age.
Risks
Not authored by or affiliated with Karpathy; it adapts his public commentary. The adoption numbers reflect the name as much as the content.
Evidence notes: Leverage: high for a single file. Maintenance: last push >2 months before fetch; single-artifact repos decay silently. Adoption: log-capped; raw stars are heavily name-driven. Eval readiness: rules are behavioral, mostly uncheckable.
Install and first run
Copy the CLAUDE.md content into your project's CLAUDE.md and prune to fit.Alternatives in this category
GitHub's first-party collection of Copilot customizations: instruction files, agent definitions, skills, and chat modes,…
A well-regarded collection of rules and knowledge files for Claude Code and Cursor. Archived by its maintainer in 2026; indexed with a…
A large library of named, reusable prompt patterns with a CLI for piping content through them. Prompt-layer infrastructure: patterns without…
Embed this badge
Maintainers of scored entries can embed their live badge; it updates when the score does and links back to this page.
[](https://agentiquette.com/index/repos/karpathy-skills)Frequently asked questions
Is multica-ai/andrej-karpathy-skills good?
multica-ai/andrej-karpathy-skills scores 61/100 (Experimental-tier) on Agentiquette's methodology v1.0, scored 2026-07-07. Strengths and weaknesses are in the editorial analysis; the score is explained dimension by dimension in the scorecard.
What is multica-ai/andrej-karpathy-skills for?
A single CLAUDE.md distilling Andrej Karpathy's public observations on LLM coding pitfalls into instruction-file rules. An instruction set, not a skill framework.
Is multica-ai/andrej-karpathy-skills maintained?
Last commit 2026-04-20, 124 open issues at fetch time (2026-07-07). Maintenance is scored at 50 of 100.
Maintainer of multica-ai/andrej-karpathy-skills? Claim this listing and suggest corrections.