andrej-karpathy-skills

andrej-karpathy-skills is a skill for Claude Code, Codex from seaworld008/Commonly-used-high-value-skills. It costs 37 tokens per session (1,734 once invoked), scanned A, original, MIT.

A set of guidelines for disciplined coding by AI agents. It emphasizes thinking through the task, making small targeted changes, stating assumptions, and checking the result.

In plain words
What is it for?
Use it for backend changes, refactoring, debugging, migrations, architecture work, or any request where scope and verification need careful control.
Why use it?
It helps prevent overbuilding, accidental changes outside the request, and unverified fixes. It is especially relevant when working in an unfamiliar or sensitive codebase.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Good fit Use it for backend changes, refactoring, debugging, migrations, architecture work, or any request where scope and verification need careful control.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/seaworld008/commonly-used-high-value-skills/andrej-karpathy-skills
Install

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.

Any agent
npx skills add seaworld008/Commonly-used-high-value-skills --skill andrej-karpathy-skills
Clone the repo
git clone --depth 1 https://github.com/seaworld008/Commonly-used-high-value-skills

Made for: Claude Code, Codex.

Wrote 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.

agentmods badge for andrej-karpathy-skills

README.md
[![agentmods](https://agentmods.dev/badge/skills/seaworld008/commonly-used-high-value-skills/andrej-karpathy-skills/github.svg)](https://agentmods.dev/skills/seaworld008/commonly-used-high-value-skills/andrej-karpathy-skills)
Your own site
<a href="https://agentmods.dev/skills/seaworld008/commonly-used-high-value-skills/andrej-karpathy-skills"><img src="https://agentmods.dev/badge/skills/seaworld008/commonly-used-high-value-skills/andrej-karpathy-skills/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.

agentmods 80×15 button for andrej-karpathy-skills

Your own site · 80×15
<a href="https://agentmods.dev/skills/seaworld008/commonly-used-high-value-skills/andrej-karpathy-skills"><img src="https://agentmods.dev/badge/skills/seaworld008/commonly-used-high-value-skills/andrej-karpathy-skills.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 37 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,734 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector warn 7 Sept 2026
SkillSpector: 2 findings, up to high

These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →

  • high Memory Poisoning · line 142
    Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.
    Fix: Protect agent memory and state from modification by untrusted content. Use read-only memory for critical instructions and validate all state changes.
  • medium Agent Snooping · line 8
    Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.
    Fix: Remove all code or instructions that list or read other skills' files or directories. Skills should operate independently; cross-skill access is a privilege escalation.
How audits are shown
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00037 $0.01734
Opus 5 $0.00018 $0.00867
Sonnet 5 $0.00007 $0.00347
Haiku 4.5 $0.00004 $0.00173

Measured 12d ago against content hash c766537eecfb, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-12, from the pricing page.

Security

Grade A, and why

andrej-karpathy-skills 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.

openclaw-skills/andrej-karpathy-skills/SKILL.md · 214 lines

How it starts

The opening of the file, as written. The whole thing — 214 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Andrej Karpathy Skills

Use this skill to keep AI-assisted engineering grounded, simple, and verifiable. It is especially useful for backend, architecture, refactoring, debugging, migrations, and any task where an agent might overbuild, guess, or change unrelated code.

The core idea is simple: do not let the model drift. Make assumptions explicit, prefer minimal changes, touch only what the request requires, and verify with concrete checks before calling the work done.

When to Use

  • Writing backend code, API changes, migrations, data access logic, or architecture-sensitive code.
  • Refactoring code where accidental behavior changes are risky.
  • Debugging a production issue or a failing test.
  • Reviewing AI-generated code for overengineering, hidden assumptions, or missing verification.
  • Implementing a user request that is ambiguous, cross-cutting, or easy to over-scope.
  • Making changes in an unfamiliar codebase.
  • Working under constraints such as dirty git state, existing style, generated files, or ownership boundaries.
  • Producing a plan before code or a concise review after code.

Skip When

  • The user explicitly asks for brainstorming without implementation.
  • The task is pure content writing, design exploration, or research without code changes.
  • A different skill with stricter domain procedure applies, such as a security audit, incident response, or regulated compliance workflow.

Core Principles

  1. Think before coding.
  2. Simplicity first.
  3. Surgical changes.
  4. Goal-driven execution.
  5. Verification before completion.
  6. Respect the existing system.
  7. Surface uncertainty early.
  8. Prefer reversible, observable steps.

Principle 1: Think Before Coding

Before implementation, establish what is known and what is assumed.

  • Read the relevant files before changing them.
  • Identify the exact behavior the user wants changed.
  • Name any ambiguity that changes the implementation path.
  • If there are multiple viable approaches, pick the smallest one that fits the repo.
  • If an approach creates a new abstraction, justify why it reduces real complexity.
  • If the task crosses module boundaries, map those boundaries first.
  • If the codebase already has a local pattern, use it.

Read the full file on GitHub · 214 lines

Changes

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.

  1. 12d ago First seen · 214 lines · 37 tokens per session scan A c766537eecfb

Subscribe to this mod's changes

andrej-karpathy-skills is a skill published in the GitHub repository seaworld008/Commonly-used-high-value-skills (70 stars, last pushed 5d ago), licensed MIT. It adds 37 tokens to every session and 1,734 once invoked, about $0.0002 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.

Related

Other skills, from other repositories

elasticsearch

Query and analyze logs in Elasticsearch. Use this skill when the user wants to search logs, query log data, count log entries, filter by trace IDs, or analyze application logs stored in Elasticsearch. Common use cases include debugging requests by trace ID, finding error logs, analyzing request patterns, and…

jawwadfirdousi/agent-skills · 68 tokens

svg-creator

Produce SVGs that are spec-correct (W3C SVG 2), CSS-independent, accessible when meaningful, safe to render in untrusted contexts, optimized in size, and readable enough to edit.

jawwadfirdousi/agent-skills · 95 tokens

prompt-template-wizard

Rigorously collects and validates all fields needed to produce a complete, unambiguous prompt template for features and bug fixes. The skill asks targeted questions until the template is fully filled, consistent, and ready to paste into a Codex/GPT-5.2 coding session.

jawwadfirdousi/agent-skills · 61 tokens

read-only-gh-pr-review

Review backend pull requests for correctness, security, performance, maintainability, and test coverage using GitHub CLI plus local repository inspection. Use when asked to review service-layer/API/database changes, audit backend branch diffs, summarize backend risk, or produce actionable must-fix/should-fix feedback.

jawwadfirdousi/agent-skills · 65 tokens

german-elster-tax-filing

Use this skill to run a complete intake for a german personal income tax return in elster for tax years 2024 onward, estimate the tax result, and map the final values into the correct official forms and fields.

jawwadfirdousi/agent-skills · 123 tokens

trello

Manage Trello boards, lists, and cards via the Trello REST API.

jawwadfirdousi/agent-skills · 19 tokens