JPeetz

MeMex Zero RAG — AI skill for Claude Code

AI community

The Karpathy LLM Wiki, production-ready.

How to install MeMex Zero RAG

This entry records only its repository, not the path inside it, so there is no exact command to give. Open JPeetz/MeMex-Zero-RAG and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What MeMex Zero RAG does

The Karpathy LLM Wiki, production-ready. Zero-RAG personal knowledge base with MCP server, multi-agent support, knowledge graph visualization, and hallucination enforcement. No embeddings, no vector DBs — just markdown + git.

Alternatives in AI

Llmwiki — Open Source Implementation of Karpathy's LLM Wiki 1.5k ★
Nocturne Memory — A lightweight, rollbackable, and visual Long-Term Memory Server for MCP Agents 1.3k ★
LLM Wiki Skill — 基于 Karpathy llm-wiki 方法论的个人知识库构建 Skill，支持多平台 1.1k ★

README

🧠 MeMex — Zero-RAG Personal Knowledge Base

[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT) [![Zero-RAG](https://img.shields.io/badge/Zero--RAG-No%20Embeddings-blue)](https://github.com/JPeetz/MeMex-Zero-RAG) [![MCP Server](https://img.shields.io/badge/MCP-Server%20Included-purple)](mcp/) [![LLM Agnostic](https://img.shields.io/badge/LLM-Agnostic-green)](https://github.com/JPeetz/MeMex-Zero-RAG) [![PRs Welcome](https://img.shields.io/badge/PRs-welcome-brightgreen.svg)](http://makeapullrequest.com)

**The Karpathy LLM Wiki pattern, production-ready.**

*No embeddings. No vector databases. No infrastructure. Just markdown, git, and your LLM.*

[Quick Start](#quick-start) • [MCP Server](mcp/) • [Prompts](PROMPTS.md) • [Schema](SCHEMA.md) • [Examples](examples/)

*"The human's job is to curate sources, direct the analysis, ask good questions, and think about what it all means. The LLM's job is everything else."* — Andrej Karpathy

Based on [Karpathy's LLM Wiki pattern](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f), extended with:

🚫 Zero hallucination enforcement — every claim must cite a source
🤖 MCP server — expose your wiki to Claude Code, Codex, OpenClaw, Cursor
👥 Multi-agent support — multiple AI agents, one shared wiki, git handles conflicts
🔒 Human-in-the-loop conflict resolution — LLM flags contradictions, you decide truth
📊 Confidence tracking — per-claim certainty scores, quarantine mode for low-confidence pages
🌐 Knowledge graph — interactive force-directed visualization of all [[wikilinks]] (optional)

Why Not RAG?

Traditional RAG retrieves document chunks every query. Your LLM rediscovers knowledge from scratch. Every. Single. Time. Nothing compounds.

**MeMex compiles knowledge once.** Your LLM reads your sources and builds a structured, interlinked wiki. Every new source makes the whole wiki richer. Knowledge grow