One-click install
npx skills add https://github.com/lovincyrus/claude-config --skill karpathy
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: karpathy
Source: https://github.com/lovincyrus/claude-config/tree/main/skills/karpathy
Command: npx skills add https://github.com/lovincyrus/claude-config --skill karpathy

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Karpathy guidelines help reduce common LLM coding mistakes by enforcing deliberate thinking, simplicity, surgical edits, and goal-driven execution across coding tasks.

Core Features & Use Cases

  • Think Before Coding: require explicit assumptions, surface tradeoffs, and ask when unclear.
  • Simplicity First: favor the smallest, most direct solution with minimal abstractions.
  • Surgical Changes: modify only what is necessary, avoiding collateral changes.
  • Goal-Driven Execution: define verifiable success criteria and loop until satisfied.

Quick Start

Apply Karpathy guidelines to your current coding task by stating assumptions, listing tradeoffs, and outlining a minimal plan before coding.

Frequently Asked Questions about karpathy

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I stop LLMs from making common coding mistakes during refactoring?

To stop LLM coding mistakes, enforce deliberate thinking by requiring explicit assumptions, favoring the smallest solution, and modifying only what is necessary. This reduces collateral changes during refactoring.

What is the best way to prompt an LLM for surgical code edits?

The best way to get surgical edits is to instruct the LLM to define verifiable success criteria and modify only what is necessary. This approach avoids collateral changes and keeps the codebase stable.

How do I apply simplicity-first coding guidelines with an LLM?

To apply simplicity-first coding guidelines, require the LLM to favor the smallest, most direct solution with minimal abstractions. State assumptions and list tradeoffs before writing any code.

Do I need a specific programming language to use think-before-code LLM guidelines?

No specific programming language is required. These LLM coding guidelines apply across any language context, from quick fixes to major refactors, ensuring goal-driven execution regardless of the stack.

Why does my LLM generate overly complex code for simple bug fixes?

LLMs generate complex code when they lack explicit constraints. By enforcing simplicity and requiring the model to surface tradeoffs and state assumptions beforehand, you ensure it chooses the most direct solution.

What are the limitations of using goal-driven execution for LLM coding tasks?

Goal-driven execution requires defining verifiable success criteria and looping until satisfied. The limitation is the overhead of structuring prompts for deliberate thinking, which may slow down trivial quick fixes.