karpathy-guidelines

Apply Karpathy-inspired behavioral guidelines to coding, reviewing, and refactoring tasks.

Updated Apr 14, 2026
One-click install
npx skills add https://github.com/gerald-ica/opencode-config-snapshot --skill karpathy-guidelines-gerald-ica
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: karpathy-guidelines
Source: https://github.com/gerald-ica/opencode-config-snapshot/tree/main/opencode/skills/karpathy-guidelines
Command: npx skills add https://github.com/gerald-ica/opencode-config-snapshot --skill karpathy-guidelines-gerald-ica

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Behavioral guidelines to reduce common LLM coding mistakes by surfacing assumptions, comparing tradeoffs, and guiding surgical changes during coding, reviewing, and refactoring.

Core Features & Use Cases

  • Think Before Coding: Explicitly state assumptions and surface uncertainties before implementing.
  • Simplicity First: Limit scope to what is asked and avoid unnecessary abstractions.
  • Surgical Changes: Modify only the necessary code, keeping other parts intact during edits.
  • Goal-Driven Execution: Define verifiable success criteria and iterate until they are met.
  • Use during code writing, code reviews, and refactoring to keep projects maintainable and safe.

Quick Start

Follow these guidelines during coding tasks to minimize overcomplication and surface assumptions.

Frequently Asked Questions about karpathy-guidelines

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

FAQPage Schema
How do I reduce LLM coding mistakes during AI-assisted development?

To reduce LLM coding mistakes, apply behavioral guidelines that surface assumptions, prioritize simplicity, and enforce surgical changes during code generation, review, and refactoring workflows.

What are the best guidelines for LLM code review and refactoring?

Effective LLM code review guidelines require thinking before coding, keeping modifications surgical, limiting scope to what is asked, and defining verifiable success criteria with clear guardrails.

Why does my LLM generate unnecessary abstractions and overcomplicate code?

LLMs overcomplicate code by ignoring scope limits; applying a simplicity-first guideline restricts changes to only what is requested, preventing unnecessary abstractions during implementation.

How do I keep LLM code changes surgical and avoid modifying unaffected code?

To keep LLM code changes surgical, enforce behavioral guidelines that explicitly restrict modifications to necessary code sections, ensuring other parts remain intact during edits and refactoring.

Can I define verifiable success criteria for LLM coding tasks?

Yes, you can define verifiable success criteria for LLM coding tasks by establishing clear guardrails and iterating through planning, implementation, and verification steps until goals are met.

When do I need behavioral guidelines for LLM-assisted coding?

You need behavioral guidelines for LLM-assisted coding when projects require maintainability and safety, specifically during code writing, code reviews, and refactoring tasks to surface uncertainties and compare tradeoffs.