Karpathy Claude File — AI skill for Claude Code
Karpathy-inspired CLAUDE.md / AGENTS.md for AI coding agents — with one critical addition: Clear is better than Clever.
How to install Karpathy Claude File
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open julianjear/karpathy-claude-file and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Karpathy Claude File does
Karpathy-inspired CLAUDE.md / AGENTS.md for AI coding agents — with one critical addition: Clear is better than Clever.
Alternatives in AI
- Impeccable — The design language that makes your AI harness better at design 13.9k ★
- Codex Plugin — 's /codex:adversarial-review command 9.6k ★
- Skill Debate — Structured four-way AI debates between Claude, Sonnet, Gemini, and Codex — use for critical decisions 2.8k ★
README
Karpathy-Inspired Coding Agent Guidelines (Clear > Clever Edition)
A `CLAUDE.md` / `AGENTS.md` file to improve AI coding agent behavior, derived from [Andrej Karpathy's observations](https://x.com/karpathy/status/2015883857489522876) on LLM coding pitfalls and modified with one critical addition: **Clear is better than Clever.**
Based on [forrestchang/andrej-karpathy-skills](https://github.com/forrestchang/andrej-karpathy-skills) — with a key principle added.
What Changed
The original guidelines are excellent but miss one of the most important software engineering principles:
**Clear is better than Clever.**
50 lines of code are not better than 100 lines of code if the 50 lines are super confusing and hard to understand.
This repo adds **"Clear Over Clever"** (Section 3) and **"Understand Before Modifying"** (Section 2) as first-class principles, and reframes "Simplicity First" to optimize for clarity *and* efficiency — not just concision.
If it's easier for a human to understand, it will likely be easier for an LLM to understand.
The Problems
From Andrej's post:
"The models make wrong assumptions on your behalf and just run along with them without checking."
"They really like to overcomplicate code and APIs, bloat abstractions... 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."
The Seven Principles
| # | Principle | Addresses |
|---|---|---|
| 1 | Don't Assume — Surface Confusion | Wrong assumptions, hidden confusion, missing tradeoffs |
| 2 | Understand Before Modifying | Changing code without understanding context or blast radius |
| 3 | Clear Over Clever | Cryptic one-liners, bad variable names, unreadable chains |
| 4 | Simplicity First | Overcomplication, bloated abstractions, speculative features |
| 5 | Surgical Changes | Orthogonal edits, drive-by refactoring |
Related Skills
Understanding Studio
An Agent Skill that turns complex topics and AI outputs into clear explanations, diagrams, interactive single-
Andrej Karpathy Skills Hermes
Hermes port of Karpathy-inspired LLM coding guidelines — SKILL.md + examples.
Andrej Karpathy Guideline For AI Coding
Andrej Karpathy-inspired guidelines for Codex and Claude Code that reduce common LLM coding mistakes through c
Codexplain
Codex explanations suck. Codexplain improves them into clear Claude-like explanations with better UX and Gordo
ASD SDE 80 Karpathy
Agent skill that makes AI answers clear: short sentences, plain words, one idea per sentence (80% ASD-STE100).
LLM Wiki Skills
Turn your markdown vault into a compounding knowledge wiki (Karpathy inspired). Six agent skills - knowledge g
Related Agents
Surgical Diff Reviewer
Reviews a diff for scope creep — every changed line must trace directly to the user's stated request. Operatio
Worker Coder
Single writer — executes boundary-clear, independent coding tasks inside one workstream (several workstreams m
Socratic Questioner
Mandatory pre-implementation interrogation. Use before any non-trivial coding task to surface assumptions, fin
