Karpathy-inspired Claude Code guidelines skill
Karpathy-Inspired Claude Code Guidelines
Four principles in one CLAUDE.md against common LLM coding mistakes: silent assumptions, overengineering, drive-by edits
- 226,019 ★ on the repository
- Last commit
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/karpathy-inspired-claude-code-guidelinesDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- Text instructions only
This is third-party code. Review the repository files before installing.
What it does
A rules file based on Andrej Karpathy's observations about model weaknesses in coding. The first principle requires stating assumptions and asking when unclear. The second forbids abstractions and features beyond the request. The third limits edits to necessary lines, and the fourth turns tasks into verifiable success criteria.
Who it is for. Developers annoyed by bloated diffs and unrequested refactoring from their agent.
Good fit when
- The agent changes code it was not asked to touch
- Solutions come out overengineered
- You want clarifying questions before implementation, not after mistakes
Not a fit when
- Tasks are trivial like typo fixes: the rules bias toward caution over speed
Example request
Add email validation to the signup form following karpathy-guidelines and leave other code aloneLimitations
These are instructions only, with no checks or hooks, so results depend on how well the model follows them. Install commands in the README point to the original forrestchang/andrej-karpathy-skills repo.
How to disable. Remove the andrej-karpathy-skills plugin or delete the section from the project's CLAUDE.md.
FAQ
How do I know the rules work?
Per the README: fewer unnecessary diff changes, fewer rewrites from overcomplication, and questions come before implementation.
Does it work with Cursor?
Yes, the repo includes a .cursor/rules/karpathy-guidelines.mdc project rule.
Discussion
No questions or reviews yet.
Related
SkillsA starter kit for Claude Code: a CLAUDE.md memory-bank template with ready hooks, subagents and skills you copy into your project
A Claude Code plugin that generates and maintains CLAUDE.md: a 0 to 100 quality score, a 150-line cap and stale reference checks
Persistent memory for Claude Code on your own database: capture of decisions and context plus a code graph, injected into every prompt
14 skills that find the gaps in a task before the code: interviews, plans, assumption tests and proof that a bug fix works
HumanLayer Skills
HumanLayer skills
Seven HumanLayer skills for Claude Code and Codex: the rpi research, design and implementation loop, visual PRs, CLAUDE.md rewrites and agentic loops
A Claude Code plugin: 39 commands for PRs, review, commits and specs, nine expert roles and hooks, in eight language versions
README in short
The README quotes Karpathy's post on LLM coding problems and offers four principles: think before coding, simplicity first, surgical changes, goal-driven execution. The rules install as a Claude Code plugin or get appended to a project CLAUDE.md. A Cursor rule is included. The author notes the rules favor caution over speed.
README badge
Original SKILL.md
--- name: karpathy-guidelines description: Behavioral guidelines to reduce common LLM coding mistakes. Use when writing, reviewing, or refactoring code to avoid overcomplication, make surgical changes, surface assumptions, and define verifiable success criteria. license: MIT --- # Karpathy Guidelines Behavioral guidelines to reduce common LLM coding mistakes, derived from [Andrej Karpathy's observations](https://x.com/karpathy/status/2015883857489522876) on LLM coding pitfalls. **Tradeoff:** These guidelines bias toward caution over speed. For trivial tasks, use judgment. ## 1. Think Before Coding **Don't assume. Don't hide confusion. Surface tradeoffs.** Before implementing: - State your assumptions explicitly. If uncertain, ask. - If multiple interpretations exist, present them - don't pick silently. - If a simpler approach exists, say so. Push back when warranted. - If something is unclear, stop. Name what's confusing. Ask. ## 2. Simplicity First **Minimum code that solves the problem. Nothing speculative.** - No features beyond what was asked. - No abstractions for single-use code. - No "flexibility" or "configurability" that wasn't requested. - No error handling for impossible scenarios. - If you write 200 lines and it could be 50, rewrite it. Ask yourself: "Would a senior engineer say this is overcomplicated?" If yes, simplify. ## 3. Surgical Changes **Touch only what you must. Clean up only your own mess.** When editing existing code: - Don't "improve" adjacent code, comments, or formatting. - Don't refactor things that aren't broken. - Match existing style, even if you'd do it differently. - If you notice unrelated dead code, mention it - don't delete it. When your changes create orphans: - Remove imports/variables/functions that YOUR changes made unused. - Don't remove pre-existing dead code unless asked. The test: Every changed line should trace directly to the user's request. ## 4. Goal-Driven Execution **Define success criteria. Loop until verified.** Transform tasks into verifiable goals: - "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: