AIBestSkill
Home/Skills/Andrej Karpathy Skills
Coding workflow · Claude Code, AI Agent Skill

Andrej Karpathy Skills: coding guardrails for Claude Code

A small Claude Code skill built around Karpathy-style coding habits: make surgical changes, state assumptions, define success criteria, and keep looping until the result is checked.

GitHub repo Adoption path Source evidence Back to skills last scan today
Job to solvecoding-agent workflow improvements

Improve code generation, review, framework-specific implementation, or developer workflow automation.

Who should use itDevelopers using AI coding agents for implementation, code review, framework work, or repo automation.

Best for developers who want a lightweight rule set they can read in a few minutes.

Works withClaude Code / AI Agent

The repo is aimed at Claude Code, but the rules are plain enough to review before adapting elsewhere.

Adoption pathSource-verified setup

The README shows a Claude Code plugin install path and a per-project CLAUDE.md option. Pick one; do not install both blindly.

Trust signalREADME + SKILL.md

1 SKILL.md + README + install evidence, 98,200 stars, and 9d ago repo freshness at last scan.

Overview

What job does it solve?Coding workflow

Improve code generation, review, framework-specific implementation, or developer workflow automation.

Who should use it?Developers using AI coding agents for implementation, code review, framework work, or repo automation.

Best for developers who want a lightweight rule set they can read in a few minutes.

Works withClaude Code / AI Agent

The repo is aimed at Claude Code, but the rules are plain enough to review before adapting elsewhere.

How can I adopt it?Source-verified setup

The README shows a Claude Code plugin install path and a per-project CLAUDE.md option. Pick one; do not install both blindly.

Can I trust it?README + SKILL.md

1 SKILL.md + README + install evidence, 98,200 stars, and 9d ago repo freshness at last scan.

What to compare?5 related skills

Compare it with broader workflow suites if you need planning, testing, or project setup skills too.

Quick read

This is the opposite of a big workflow pack. It is one skill file focused on how the agent should behave while writing or reviewing code.

Use it when your agent tends to overbuild, skip verification, or hide assumptions.

Why we list it
  • Public because this is a source-backed installable skill-like repo with enough current evidence to qualify for the shortlist.
  • Current GitHub metadata is available for forrestchang/andrej-karpathy-skills.
  • 1 SKILL.md file(s) were found in the recursive tree scan.
  • Trusted install evidence was extracted from 1 source file(s).

Adoption path for Andrej Karpathy Skills

Primary adoption path

The README gives both plugin and per-project setup paths. Choose the one that matches how you use Claude Code.

  • /plugin marketplace add forrestchang/andrej-karpathy-skills

Source evidence

These excerpts show the repo's own claims and setup paths. Read them before deciding how to install it.

README excerpt

README.md

The Problems

From Andrej's post:

"The models make wrong assumptions on your behalf and just run along with them without checking. They don't manage their confusion, don't seek clarifications, don't surface inconsistencies, don't present tradeoffs, don't push back when they should."
"They really like to overcomplicate code and APIs, bloat abstractions, don't clean up dead code... implement a bloated construction over 1000 lines when 100 would do."
"They still sometimes change/remove comments and code they don't sufficiently understand as side effects, even if orthogonal to the task."

The Solution

Four principles in one file that directly address these issues:

PrincipleAddresses
Think Before CodingWrong assumptions, hidden confusion, missing tradeoffs
Simplicity FirstOvercomplication, bloated abstractions
Surgical ChangesOrthogonal edits, touching code you shouldn't
Goal-Driven ExecutionLeverage through tests-first, verifiable success criteria
4. Goal-Driven Execution

Define success criteria. Loop until verified.

Transform imperative tasks into verifiable goals:

Instead of...Transform to...
"Add validation""Write tests for invalid inputs, then make them pass"
"Fix the bug""Write a test that reproduces it, then make it pass"
"Refactor X""Ensure tests pass before and after"

For multi-step tasks, state a brief plan:

1. [Step] → verify: [check]
2. [Step] → verify: [check]
3. [Step] → verify: [check]

Strong success criteria let the LLM loop independently. Weak criteria ("make it work") require constant clarification.

Install

Option A: Claude Code Plugin (recommended)

From within Claude Code, first add the marketplace:

/plugin marketplace add forrestchang/andrej-karpathy-skills

Then install the plugin:

/plugin install andrej-karpathy-skills@karpathy-skills

This installs the guidelines as a Claude Code plugin, making the skill available across all your projects.

Option B: CLAUDE.md (per-project)

New project:

curl -o CLAUDE.md https://raw.githubusercontent.com/forrestchang/andrej-karpathy-skills/main/CLAUDE.md

Existing project (append):

echo "" >> CLAUDE.md
curl https://raw.githubusercontent.com/forrestchang/andrej-karpathy-skills/main/CLAUDE.md >> CLAUDE.md

Customization

These guidelines are designed to be merged with project-specific instructions. Add them to your existing CLAUDE.md or create a new one.

For project-specific rules, add sections like:

Show scanned SKILL.md paths (1)
  • skills/karpathy-guidelines/SKILL.md