karpathy-guidelines

Apply Karpathy-style behavioral guidelines to reduce LLM coding mistakes.

Updated Mar 17, 2026
One-click install
npx skills add https://github.com/Blake-John/agent-config --skill karpathy-guidelines-blake-john
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: karpathy-guidelines
Source: https://github.com/Blake-John/agent-config/tree/main/.agents/skills/general-principle
Command: npx skills add https://github.com/Blake-John/agent-config --skill karpathy-guidelines-blake-john

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Behavioral guidelines to reduce common LLM coding mistakes, derived from Andrej Karpathy's observations on LLM coding pitfalls. The guidelines help teams avoid overcomplication, surface assumptions, and define verifiable success criteria.

Core Features & Use Cases

  • Think-before-code discipline: state assumptions, surface tradeoffs, and surface uncertainties.
  • Simplicity-first rule: limit features and avoid unnecessary abstractions.
  • Surgical changes: modify only what's necessary and avoid touching unrelated code.

Quick Start

Apply these guidelines whenever you write, review, or refactor code to keep changes surgical, surface assumptions, and define verifiable success criteria.

Frequently Asked Questions about karpathy-guidelines

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

FAQPage Schema
How do I reduce common LLM coding mistakes during code generation?

To reduce common LLM coding mistakes, apply behavioral guidelines that enforce explicit assumptions, surface tradeoffs, and define verifiable success criteria before writing code. This keeps changes surgical and well-scoped.

What are the best behavioral guidelines for keeping code changes surgical and minimal?

The best guidelines for surgical code changes emphasize a simplicity-first rule to avoid unnecessary abstractions, require stating assumptions explicitly, and mandate modifying only what is necessary without touching unrelated code.

How do I enforce verifiable success criteria when refactoring code?

To enforce verifiable success criteria when refactoring, apply think-before-code discipline to surface uncertainties and tradeoffs upfront, ensuring all modifications remain minimal and functionally verifiable.

Does this approach to code review work for general software projects?

Yes, this approach to code review works for general software projects by applying behavioral guidelines during writing, reviewing, and refactoring to limit features and avoid overcomplication across any codebase.

Why should I surface tradeoffs and assumptions before writing new code?

You should surface tradeoffs and assumptions before writing new code to establish think-before-code discipline, which prevents overcomplication and ensures that all subsequent changes remain well-scoped and verifiable.