OpenRuna
Sign in

karpathy-guidelines

TEXT
Prompt2417 chars · 355 words
---
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:
\
Strong success criteria let you loop independently. Weak criteria ("make it work") require constant clarification.

Overview

karpathy-guidelines is a free prompt on OpenRuna. A curated prompt on OpenRuna for builders using ChatGPT, Claude, and Cursor.

What this prompt does

"karpathy-guidelines" is designed to help you get reliable results from AI assistants for real prompt tasks. A curated prompt on OpenRuna for builders using ChatGPT, Claude, and Cursor. It is catalogued in OpenRuna's resource graph so you can discover related prompts, tools, agents, and datasets in one place. Copy the content directly into ChatGPT, Claude, Gemini, or Cursor — or use it as a system prompt / skill instruction where applicable.

Use cases

  • Use "karpathy-guidelines" when you need a repeatable prompt for professional workflows without writing instructions from scratch each time.
  • Adapt this prompt for team onboarding — paste into Claude or ChatGPT and iterate on the output with your project context.
  • Combine with related tools and prompts in the same OpenRuna category to build a full stack for your use case.
  • Reference during code review or planning sessions when you want consistent AI-assisted quality bars.

Example output

When you run this prompt, expect structured output similar to:

---
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 Befor…

Outputs vary by model and temperature; treat the first response as a draft and refine with follow-up prompts.

Tips by platform

Claude

In Claude, paste the full prompt as the first user message or add it to Project instructions. Ask Claude to confirm assumptions before executing. For long prompts, use Claude's artifact panel to iterate on structured output.

ChatGPT

In ChatGPT, start a new chat and paste this prompt verbatim. Enable GPT-4o or your preferred model for coding tasks. Use follow-ups like "apply this to [your context]" for best results.

Cursor

In Cursor, add key instructions from this prompt to .cursorrules or a SKILL.md file. Reference it in Agent mode with @ mentions. Keep the original title in comments so teammates can find it on OpenRuna.

Frequently asked questions

What is "karpathy-guidelines"?
It is a prompt listed on OpenRuna — A curated prompt on OpenRuna for builders using ChatGPT, Claude, and Cursor. You can copy and adapt it for ChatGPT, Claude, Cursor, or other AI tools.
Is this prompt free to use?
Most OpenRuna resources are open or CC0-licensed. Check the license on this page before commercial use. Premium collections are clearly marked.
How do I get the best results?
Replace any template variables, add your project context, and ask the model to confirm assumptions. Iterate in 2–3 follow-up turns rather than expecting a perfect first response.
Can I use this with Claude and ChatGPT?
Yes. The prompt is model-agnostic text. Tips on this page cover Claude, ChatGPT, and Cursor specifically.
Where can I find related resources?
Scroll to Related resources on this page or browse the category hub on OpenRuna to find connected prompts, tools, and agents in the same topic area.

Related resources