# Karpathy Guidelines

> 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.

Source: https://skillsagentes.com/skills/multica-ai/andrej-karpathy-skills/karpathy-guidelines
Repository: https://github.com/multica-ai/andrej-karpathy-skills
Author: multica-ai
License: MIT
Updated: hace 6 meses
Context cost: 55 tok installed, 630 tok once triggered, 630 tok with every bundled file
Bundle: 1 file, 2 KB
Permissions requested: none declared

## Install

```bash
npx -y skills add multica-ai/andrej-karpathy-skills --skill karpathy-guidelines --agent claude-code
```

## What it does

- Makes the agent state its assumptions before it writes code, and stop and name what is confusing instead of silently picking one reading of an ambiguous request.
- Caps scope to the minimum that solves the problem: no speculative features, no abstractions for single-use code, no error handling for impossible cases.
- Keeps edits surgical — no reformatting or refactoring of adjacent code, matching existing style even where the agent would choose differently, and cleaning up only the orphans its own change created.
- Turns vague tasks into verifiable goals: "add validation" becomes "write tests for invalid inputs, then make them pass", with a numbered plan carrying a check per step.

## Use it when

- Writing, reviewing, or refactoring code with an agent that tends to overbuild.
- Working in an existing codebase where unrequested refactors and reformatting turn a small change into a large diff.
- Handing over a loosely specified task and wanting explicit success criteria before any code is written.

## Don't bother when

- The task is trivial — the file says outright that these guidelines bias toward caution over speed, and to use judgment there.
- You want the agent to keep moving without pausing; this deliberately makes it stop and surface confusion instead.

## What triggers it

- "Before you write any code, tell me your assumptions"
- "Refactor this module but keep the change surgical"
- "Add validation here and define how we will verify it"
- "Review this diff for overcomplication"

## Before you install

- None. One markdown file with no scripts, no binaries and no environment variables — it changes how the agent behaves, not what it can reach.

## Files

- SKILL.md — 2 KB

## SKILL.md

Reproduced verbatim from multica-ai/andrej-karpathy-skills under MIT. This section is the upstream document and is in English.

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

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

---

Skills Agentes — https://skillsagentes.com/skills/multica-ai/andrej-karpathy-skills/karpathy-guidelines