A single CLAUDE.md file distilling Andrej Karpathy's observations on LLM coding pitfalls into 4 hard rules: think before coding, prefer simplicity over abstraction, make targeted changes only, and verify by outcome not intent. Packaged as a proper Agent Skill with .cursor/rules, plugin.json, and multi-tool support. 140K stars β the fastest validation that developers already knew AI was doing these things wrong.
What Karpathy Guidelines (multica-ai) can do for your AI workflow
Single claude directly from your Cursor conversation
Works across Cursor, Claude Code, Codex CLI β install once, use everywhere
Supports karpathy, best-practices, claude-md, anti-patterns and more
One-command installation β no complex setup required
Combine with other skills to build powerful multi-step AI workflows
Try these prompts with your AI agent after installing Karpathy Guidelines (multica-ai)
Help me get started with Karpathy Guidelines (multica-ai)
Explains what Karpathy Guidelines (multica-ai) does, walks through the setup, and runs a quick demo based on your current project
Use Karpathy Guidelines (multica-ai) to a single CLAUDE
Invokes Karpathy Guidelines (multica-ai) with the right parameters and returns the result directly in the conversation
What can I do with Karpathy Guidelines (multica-ai) in my ai agent workflow?
Lists the top use cases for Karpathy Guidelines (multica-ai), with example commands for each scenario
Guides & tutorials for AI skills
10 Fresh GitHub Agent Skills Added in June 2026
A hand-checked shortlist of recently updated GitHub agent skills that were not already in our catalog, excluding marketplaces, awesome lists, managers, and generic skill directories.
The 7 AI Skills Every Software Developer Should Have Installed in 2026
After testing dozens of developer-focused AI skills, these are the seven that have proven genuinely useful across different tech stacks and workflows β not just impressive demos, but tools that hold up under daily use.
Karpathy Guidelines (multica-ai) extends your AI assistant with the ability to a single CLAUDE.md file distilling Andrej Karpathy's observations on LLM coding pitfalls into 4 hard rules: think before coding, prefer simplicity over abstraction, make targeted changes only, and verify by outcome not intent. Packaged as a proper Agent Skill with .cursor/rules, plugin.json, and multi-tool support. 140K stars β the fastest validation that developers already knew AI was doing these things wrong. Rather than leaving your conversation to handle this manually, you can ask your Cursor agent directly β and it will take care of the task end-to-end, using Karpathy Guidelines (multica-ai) as its underlying capability.
Karpathy Guidelines (multica-ai) works across Cursor, Claude Code, Codex CLI through the Model Context Protocol (MCP) β an open standard that lets AI clients share tools and skills without lock-in. This makes it especially practical for teams and individuals working with karpathy and best-practices. Because MCP is platform-agnostic by design, you install Karpathy Guidelines (multica-ai) once and it becomes available across all your AI clients. Whether you're working in Cursor for focused sessions or Claude Code for integrated workflows, the skill behaves consistently.
To install Karpathy Guidelines (multica-ai), copy the skill folder to `~/.claude/skills/multica-karpathy-guidelines/` for use across all your projects, or `.claude/skills/multica-karpathy-guidelines/` for a single project. Restart Cursor and the skill is immediately active β invoke it with `/multica-karpathy-guidelines` or just describe your goal and the AI will pick it up automatically.
Karpathy Guidelines (multica-ai) is a community-contributed AI productivity skill available through DiscoverAISkills. The source code is open on GitHub β you can inspect it, contribute fixes, or fork it to adapt the skill for your specific setup. Like all skills on DiscoverAISkills, it is free to install and use. The broader AI skills ecosystem continues to expand as developers contribute new capabilities across categories like developer tools, data analysis, writing, automation, and more.
Place the skill folder at ~/.claude/skills/multica-karpathy-guidelines/ for personal use (all projects), or .claude/skills/multica-karpathy-guidelines/ for project-specific use. Restart your AI client, then invoke with /multica-karpathy-guidelines or let the AI discover it automatically.
Karpathy Guidelines (multica-ai) supports Cursor, Claude Code, Codex CLI. It integrates seamlessly with these AI platforms to extend their capabilities.
Karpathy Guidelines (multica-ai) is free to install. Check the repository for licensing information.
A single CLAUDE.md file distilling Andrej Karpathy's observations on LLM coding pitfalls into 4 hard rules: think before coding, prefer simplicity over abstraction, make targeted changes only, and verify by outcome not intent. Packaged as a proper Agent Skill with .cursor/rules, plugin.json, and multi-tool support. 140K stars β the fastest validation that developers already knew AI was doing these things wrong.
Caveman
A radical simplicity skill that strips AI coding assistants back to first principles. Caveman enforces brutally simple code β no abstractions, no frameworks, no clever tricks. Just plain, readable, maintainable code that any developer can understand at a glance. Inspired by the philosophy that the best code is the code you don't write.
Karpathy Guidelines
A comprehensive collection of AI coding guidelines curated from Andrej Karpathy's public writings, talks, and social media posts. Distills the legendary AI researcher's hard-won wisdom on building reliable AI systems, debugging neural networks, and writing clean ML code into an actionable skill for Claude Code and other AI assistants.
Awesome Claude Code (Community Edition)
karpathy-llm-wiki
Persistent wiki manager based on Karpathy's LLM-Wiki pattern. Builds and maintains a structured, interlinked markdown wiki from your sources β compounding kn...
guidelines
Automatically generate HTML preview pages for list pages, form pages and details pages that comply with FTdesign specifications based on the user's natural language description, enabling direct preview in the browser.
Karpathy Curated RSS Brief
Fetch articles from Karpathy's curated 93 RSS feeds and generate a Chinese tech daily newsletter. Triggers: RSS daily, RSS newsletter, karpathy-curated-rss-brief, daily newsletter, Kar...
Select your agent
Option 1: Install via CLI (recommended)
Recommended (no pre-install needed)
npx clawhub@latest --dir ~/.claude/skills install multica-karpathy-guidelinesOr via clawhub CLI (if already installed)
clawhub --dir ~/.claude/skills install multica-karpathy-guidelinesβ οΈ Requires Node.js 18+. No Node? Use Option 2 below to download the ZIP instead. Install Node.js β
Option 2: Manual install (no Node required)
Download the ZIP, extract it, and place the folder at the path below. Restart your agent to activate.
Install path
~/.claude/skills/multica-karpathy-guidelines/π‘Extract and place the folder at the path above, then restart your agent.
I need to work with karpathy and best-practices
Recognizes the task, selects Karpathy Guidelines (multica-ai) as the right tool, and completes it without leaving the chat
Automate my ai agent tasks using Karpathy Guidelines (multica-ai)
Identifies repetitive steps in your workflow and sets up Karpathy Guidelines (multica-ai) to handle them automatically
The community's most comprehensive Claude Code resource guide. 86+ real-world tips, tricks, and best practices collected from developers who use Claude Code daily β covering skills, agents, hooks, commands, and workflows. Unlike official documentation, this is battle-tested knowledge from practitioners: what actually works, what to avoid, and how to get the most out of Claude Code on complex projects. Actively maintained with contributions from engineers at major tech companies.