Andrej Karpathy

Based on Andrej Karpathy's observations about common LLM coding pitfalls, this skill teaches Claude to surface assumptions before coding, avoid speculative abstractions, make surgical changes instead of sprawling refactors, and define verifiable success criteria for every task. It's designed to counter the tendency of LLMs to overcomplicate solutions and touch more code than necessary. The tradeoff is clear: it biases toward caution over speed, so you'll want to skip it for trivial tasks. Best used when you're reviewing code, making changes to existing systems, or dealing with anything where "just make it work" has burned you before.

install copies · 90 days. Times the install command was copied on this site in the last 90 days, bots filtered.: 12
skills.sh installs: 193
GitHub stars: 39.4K
last updated: Jun 3, 2026

View on GitHub

Install

$ npx -y skills add sickn33/antigravity-awesome-skills --skill andrej-karpathy --agent claude-code

Installs into .claude/skills of the current project.

Files

Karpathy Guidelines

Behavioral guidelines to reduce common LLM coding mistakes, derived from Andrej Karpathy's observations on LLM coding pitfalls.

Tradeoff: These guidelines bias toward caution over speed. For trivial tasks, use judgment.

When to Use This Skill

Use when writing, reviewing, or refactoring code with an LLM.
Use when a change needs to stay surgical and avoid speculative abstractions.
Use when assumptions, tradeoffs, and verification criteria should be made explicit.
Use when code has become overcomplicated and needs to be simplified.