Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add LearnPrompt/andrej-karpathy-skills --skill karpathy-idea-filesgit clone --depth 1 https://github.com/LearnPrompt/andrej-karpathy-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/learnprompt/andrej-karpathy-skills/karpathy-idea-files)<a href="https://agentmods.dev/skills/learnprompt/andrej-karpathy-skills/karpathy-idea-files"><img src="https://agentmods.dev/badge/skills/learnprompt/andrej-karpathy-skills/karpathy-idea-files/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/learnprompt/andrej-karpathy-skills/karpathy-idea-files"><img src="https://agentmods.dev/badge/skills/learnprompt/andrej-karpathy-skills/karpathy-idea-files.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00114 | $0.01565 |
| Opus 5 | $0.00057 | $0.00783 |
| Sonnet 5 | $0.00023 | $0.00313 |
| Haiku 4.5 | $0.00011 | $0.00156 |
Grade A, and why
karpathy-idea-files scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 192 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Skill 9: Idea Files / Gist-First Sharing(想法文件优先)
Source: https://x.com/karpathy/status/2040470801506541998 | https://github.com/multica-ai/andrej-karpathy-skills "LLM Knowledge Bases idea file" — 27k likes
Core Principle
Ship the idea, not the implementation.
Karpathy's Gist pattern: write a clear abstract + success criteria + constraints, then let an agent (or anyone) implement it for their specific context. One idea spec → infinite implementations.
The old model: write code → share code → others adapt code. The Karpathy model: write idea spec → share spec → agents implement for context.
The Idea File Template
Every new idea gets this format before any code is written:
# [IDEA NAME]
> Status: [DRAFT / SPEC / ACTIVE / ARCHIVED]
> Created: [DATE]
> Source inspiration: [LINK or "original"]
## Abstract (2-3 sentences)
What is this? What problem does it solve? Why does it matter?
## The Core Insight
What's the one non-obvious observation that makes this idea work?
(If you can't write this, the idea isn't ready yet.)
## Success Criteria
When is this "done"? List 3-5 verifiable outcomes:
- [ ] [Observable outcome 1]
- [ ] [Observable outcome 2]
- [ ] [Observable outcome 3]
## Minimal Viable Version
Describe the simplest possible thing that would prove the concept works.
(One paragraph. No code yet.)
## Constraints
- Must work with: [tools/stack/dependencies]
- Must NOT require: [what to avoid]
- Scale target: [personal use / team / public]
- Time budget: [hours/days]
## Open Questions
Things I don't know yet that matter:
1. [Question 1]
2. [Question 2]
## Implementation Notes (optional)
Hints for whoever (agent or human) implements this:
- Suggested approach: [rough approach]
- Key risk: [what might go wrong]
- Prior art: [similar things that exist]
## Example Input/Output
Input: [example]
Output: [example]
Converting a Vague Idea to a Spec
When you have a rough idea, use this prompt to crystallize it:
I have a rough idea I want to turn into an implementation-ready spec.
Rough idea: [YOUR VAGUE IDEA IN NATURAL LANGUAGE]
Help me write a proper idea file with:
1. A precise 2-sentence abstract (what + why)
2. The core non-obvious insight (the thing that makes it work)
3. 3-5 verifiable success criteria
4. The minimal viable version (simplest proof of concept)
5. 3 things an implementer needs to know
6. 2 open questions I should resolve before building
Be precise. Remove vagueness. If something is unclear, ask — don't invent.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 192 lines · 114 tokens per session scan A 1821a5028852
karpathy-idea-files is a skill published in the GitHub repository LearnPrompt/andrej-karpathy-skills (97 stars, last pushed 2mo ago), licensed MIT. It adds 114 tokens to every session and 1,565 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
local-ai-agents
Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…
next-cache-components-adoption
Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…
insight-error-page
Write or audit an insight-kind error page for the Next.js dev overlay. Use when creating a new errors/ .mdx page, auditing an existing one, or checking that a page matches the framework fix cards. Covers page structure, title alignment, FixCard cards with Copy prompt button, code snippets, terminology verification…
next-cache-components-optimizer
Drive a Next.js route to instant navigation by setting up an agentic loop, under Cache Components / PPR, on initial load (hard navigation) and client-side navigation (soft navigation). Encode the goal as a failing @next/playwright instant() e2e and work it to green, one verified route at a time; the shipped test then…
next-partial-prefetching-adoption
Turn on Partial Prefetching in a Next.js app and work through the insights it surfaces. Use when the user wants to enable or adopt Partial Prefetching, flip the partialPrefetching flag, opt routes in with export const prefetch = 'partial', audit Link prefetch={true} behavior, preserve existing prefetched UI with…