Skip to main content

Karpathy Guidelines

Skill Verified Active

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.

Purpose

To improve the quality and reduce the common mistakes made by LLMs when writing, reviewing, or refactoring code.

Features

  • Provides explicit principles for LLM code generation
  • Encourages stating assumptions and surfacing tradeoffs
  • Promotes simple, focused code without speculation
  • Ensures changes are surgical and limited to the task scope
  • Defines success criteria for verifiable execution

Use Cases

  • When writing new code to ensure adherence to best practices
  • During code reviews to identify potential LLM-induced mistakes
  • When refactoring code to maintain focus and avoid unnecessary changes
  • To guide LLM interactions for more predictable and high-quality code output

Non-Goals

  • Automatically fixing all LLM coding mistakes
  • Replacing human oversight in the coding process
  • Providing project-specific coding standards
  • Performing code execution or validation itself

Installation

First, add the marketplace

/plugin marketplace add forrestchang/andrej-karpathy-skills
/plugin install andrej-karpathy-skills@karpathy-skills

Quality Score

Verified
99 /100
Analyzed about 23 hours ago

Trust Signals

Last commit24 days ago
Stars128.1k
Status
View Source

© 2025 SkillRepo · Find the right skill, skip the noise.