kguide

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

Updated Sep 7, 2024
One-click install
npx skills add https://github.com/Asylcreek/home-manager --skill kguide
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: kguide
Source: https://github.com/Asylcreek/home-manager/tree/main/dots/agents/skills/kguide
Command: npx skills add https://github.com/Asylcreek/home-manager --skill kguide

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Behavioral guidelines to reduce common LLM coding mistakes, derived from Karpathy's observations on LLM coding pitfalls, and designed to be used during writing, reviewing, or refactoring code.

Core Features & Use Cases

  • Think before coding: surface assumptions, state tradeoffs, and ask for clarification.
  • Simplicity first: deliver minimum, well-scoped code with no unnecessary abstractions.
  • Surgical changes: modify only the necessary parts and clean up after.
  • Goal-driven execution: translate tasks into measurable goals and verification criteria.

Quick Start

State your assumptions explicitly, surface tradeoffs, and ask for clarification before coding.

Frequently Asked Questions about kguide

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

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

Reduce LLM coding mistakes during refactoring by applying behavioral guidelines that enforce simplicity, surface assumptions, and restrict edits to only necessary parts. This ensures surgical changes without overcomplication.

What are the best practices for LLM code review and writing new code?

Best practices for LLM code review involve stating assumptions explicitly, translating tasks into measurable goals, and defining verifiable success criteria before execution. This prevents unnecessary abstractions and maintains well-scoped code.

How does a plan-driven execution model improve LLM coding outcomes?

A plan-driven execution model improves LLM coding outcomes by enforcing explicit steps, checklists, and measurable criteria. It forces the LLM to think before coding, state tradeoffs, and ask for clarification.

Why does my LLM generate overcomplicated code with unnecessary abstractions?

LLMs generate overcomplicated code with unnecessary abstractions when not constrained by simplicity-first guidelines. Enforcing surgical changes and minimum well-scoped code delivery prevents the LLM from introducing unneeded complexity.

Can I use Karpathy-inspired coding guidelines for refactoring existing code?

Yes, you can use Karpathy-inspired coding guidelines for refactoring existing code. They dictate modifying only necessary parts, cleaning up afterwards, and ensuring changes remain surgical and goal-driven.