skills.rest
Flowm
FlowmCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

karpathy-guidelines

Enforce minimal, surgical code edits and explicit assumptions during LLM coding tasks.

3|1|Updated Nov 5, 2015
Search Tags:#code quality#code review#refactoring#software development#error prevention#coding guidelines#llm coding
One-click install
npx skills add https://github.com/Flowm/dotfiles --skill karpathy-guidelines-flowm
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: karpathy-guidelines
Source: https://github.com/Flowm/dotfiles/tree/main/home/.agents/skills/karpathy-guidelines
Command: npx skills add https://github.com/Flowm/dotfiles --skill karpathy-guidelines-flowm

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill addresses the frequent issue of LLMs producing overcomplicated, unnecessary, or error-prone code during writing, review, and refactoring tasks, which leads to wasted time, technical debt, and avoidable bugs.

Core Features & Use Cases

  • Pre-Coding Assumption Surfacing: Requires explicit stating of assumptions and tradeoffs before writing code, eliminating hidden confusion and misalignment.
  • Simplicity-First Development: Enforces minimal, request-aligned code with no speculative features, unnecessary abstractions, or unrequested error handling.
  • Surgical Code Edits: Mandates that changes only touch requested code, match existing style, and avoid unrequested refactors of unrelated code.
  • Use Case: When refactoring a legacy user authentication module, use this Skill to avoid adding unrequested features, only modify the required functions, and define clear test criteria to verify the refactor doesn't break existing login flows.

Quick Start

Use the karpathy-guidelines skill to review your recent pull request changes for overcomplication and unnecessary edits before merging.

Frequently Asked Questions about karpathy-guidelines

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

FAQPage Schema
How do I stop LLMs from making overcomplicated code changes during refactoring?▼

To prevent overcomplicated code changes during refactoring, enforce strict guidelines for minimal viable code delivery and surgical edits. This ensures modifications only touch requested code, match existing styles, and avoid unrequested refactors of unrelated modules to eliminate technical debt.

What are the best coding guidelines to reduce LLM coding mistakes in pull request reviews?▼

Effective coding guidelines for pull request reviews mandate pre-implementation assumption surfacing and verifiable success criteria definition. This process eliminates hidden confusion, prevents unnecessary abstractions, and ensures request-aligned code delivery before merging changes to eliminate rework.

Why does my LLM add unrequested features and error handling when implementing new functionality?▼

LLMs add unrequested features and error handling due to a lack of simplicity-first development constraints. Enforcing strict behavioral guidelines ensures the AI delivers minimal, request-aligned code without speculative features or unnecessary abstractions, preventing avoidable bugs and wasted time.

How do I surface hidden assumptions before writing code with an LLM?▼

Surface hidden assumptions by enforcing a pre-coding phase that requires explicit stating of assumptions and tradeoffs. This practice eliminates hidden confusion and misalignment before any code is written, ensuring software development aligns with actual requirements and reduces rework.

Can I use these coding guidelines for legacy code refactoring and bug fix delivery?▼

Yes, you can use these coding guidelines for legacy code refactoring and bug fix delivery. They enforce surgical edit practices and verifiable success criteria to ensure changes do not break existing flows, eliminate technical debt, and prevent avoidable bugs during software development.

Related Skills

Explore More Software Engineering Skills→
JamesClonkJamesClonk

karpathy-guidelines

Apply Karpathy guidelines to coding tasks for explicit assumptions and verifiable steps.

Community
Intermediate
ThejanaJayalathThejanaJayalath

ai-regression-testing

Automate regression testing for AI-assisted development with Vitest and Next.js.

Community
Advanced
nicolanoviellnicolanoviell
2

code-review

Review code changes for bugs, style issues, and best practices.

Community
Intermediate
mathur-exemathur-exe

improve-prompt-cev

Refine prompts to steer AI behavior toward coherent extrapolated volition.

Community
Basic