agentleFS
ntroccolintroccoli/karpathy-ideas/.claude/skills/karpathy-ideas/SKILL.md

karpathy-ideas

Skill1 starsChanged 2 days agoMITView on GitHub

Look up Andrej Karpathy's public ideas on working with LLMs and AI agents (agentic coding, verification loops, context, output formats, knowledge bases, agent-native software) and apply them to the current task. Use when designing an agent workflow, a prompt, a coding process, a knowledge base or a tool for agents, or when the user asks what Karpathy recommends.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Prefer status: current. If an idea is superseded, follow superseded_by to his newer view.

    line 26
  2. Distinctive:

    Do not reproduce his words beyond the quote in the file.

    line 31

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Karpathy Ideas

A curated, unofficial collection of Karpathy's public ideas, one file per idea, with date, source link, summary, at most one short quote, and Apply steps.

Location: the repository root containing this skill (index.md, ideas/, themes/). If the skill was copied elsewhere, read https://raw.githubusercontent.com/ntroccoli/karpathy-ideas/main/llms.txt and follow it.

How to use it

For a whole project, follow "Applying the ideas to a project" in llms.txt: a project with a record (docs/karpathy-ideas.md) gets a revisit, which reads corrections.md first; one without gets a first review of checklist.md. Answer each question (measure where it says "Ask"), then open only the linked ideas that point to a gap. For one task:

  1. Delegate the search to a cheaper subagent (for example Haiku or Sonnet) when you can: it reads the 2-3 matching pages in themes/ (or index.md for everything) and returns a shortlist of idea files with one line each on why they fit.
  2. Open only the 2-5 shortlisted idea files.
  3. Prefer status: current. If an idea is superseded, follow superseded_by to his newer view.
  4. Ideas marked "Anthropic period" date from after he joined Anthropic (2026-05-19) and concern Claude or Anthropic; mention that when you rely on them.
  5. Turn the Apply steps into concrete actions for the user's situation. Cite each idea by title and source link.
  6. Do not reproduce his words beyond the quote in the file; link to the source.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.