LLM Wiki
LLM-powered knowledge base from your Claude Code, Codex CLI, Copilot, Cursor & Gemini sessions. Karpathy's LLM Wiki pattern — implemented and shipped.
What is LLM Wiki?
LLM Wiki is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to llm-powered knowledge base from your claude code, codex cli, copilot, cursor & gemini sessions. karpathy's llm wiki pattern — implemented and shipped.
LLM-powered knowledge base from your Claude Code, Codex CLI, Copilot, Cursor & Gemini sessions. Karpathy's LLM Wiki pattern — implemented and shipped.
This server falls under the Knowledge & Memory category on MCPgee, the world's largest MCP server directory with 33,000+ servers.
Features
- LLM-powered knowledge base from your Claude Code, Codex CLI,
Use Cases
Maintainer
Works with
Installation
Manual Installation
npx llm-wikiConfiguration
Configuration Details
claude_desktop_config.json
Performance
Response Metrics
Resource Usage
How to Set Up and Use LLM Wiki
llmwiki is a command-line tool that transforms the dormant session transcripts produced by Claude Code, Codex CLI, GitHub Copilot, Cursor, Gemini CLI, and Obsidian into a searchable, interlinked knowledge base — implementing Andrej Karpathy's LLM Wiki pattern. It converts raw `.jsonl` session files into a Karpathy-style wiki with structured pages for sources, entities, concepts, syntheses, and comparisons, then builds a beautiful static site with global search, dark mode, and activity heatmaps. The wiki also produces AI-consumable exports (`llms.txt`, `llms-full.txt`, JSON-LD graph, per-page `.json` siblings) so other AI agents can query your accumulated knowledge directly.
Prerequisites
- Python 3.9 or later installed
- Session transcript files from at least one supported tool: Claude Code, Codex CLI, GitHub Copilot, Cursor, or Gemini CLI (raw .jsonl files on disk)
- Git (for cloning the repository)
- An MCP-compatible client if using the MCP server interface
Clone the llmwiki repository
Clone the repository to your local machine. The project is self-contained with a setup script that handles Python dependency installation.
git clone https://github.com/Pratiyush/llm-wiki && cd llm-wikiRun the one-time setup script
Run `setup.sh` to install Python dependencies and configure the project for your environment. This only needs to be done once.
./setup.shPoint llmwiki at your session transcripts
Place your raw agent session files (`.jsonl` format) into the `raw/` directory. This directory is gitignored to protect your personal data. Each file becomes a session in the wiki.
# Example: copy Claude Code sessions
cp ~/.claude/projects/*/*.jsonl raw/Build the wiki
Run the build script to convert raw session files into structured markdown wiki pages, generate the static site, and produce AI-consumable exports.
./build.shServe and browse the wiki locally
Start the local development server. The wiki will be available at http://127.0.0.1:8765 with full search, dark mode, keyboard shortcuts, and session filtering.
./serve.sh
# Open http://127.0.0.1:8765 in your browserConfigure as an MCP server for AI agent access
Register llmwiki as an MCP server so AI assistants can query your knowledge base directly during sessions.
LLM Wiki Examples
Client configuration
Add llmwiki as an MCP server in your claude_desktop_config.json so Claude can query your session knowledge base.
{
"mcpServers": {
"llm-wiki": {
"command": "npx",
"args": ["llm-wiki"]
}
}
}Prompts to try
Once your wiki is built and connected, use these prompts to query your accumulated session knowledge.
- "What have I learned about rate limiting across my past sessions?"
- "Show me all sessions where I worked on authentication"
- "Summarize the entities and concepts from my Claude Code sessions this month"
- "What questions from past sessions are still unresolved?"
- "Compare how I approached API design in different projects"Troubleshooting LLM Wiki
The build script fails with 'No sessions found'
Ensure your raw session files are placed in the `raw/` directory at the project root and have the `.jsonl` extension. The build script looks specifically in this directory. You can use the example demo sessions in `examples/demo-sessions/` to verify the build works before adding your own data.
llms.txt or graph.jsonld are not generated after build
These AI-consumable exports require a successful full build. Check the build output for errors. If specific pages fail lint checks (16 structural and LLM-powered rules are enforced), they may be excluded from exports. Run `./build.sh` with verbose output to identify failing pages.
The static site search does not return results
The Cmd+K command palette uses a pre-built fuzzy search index generated during `./build.sh`. If search returns nothing, re-run the build to regenerate the index, then refresh the page. Ensure JavaScript is enabled in your browser as the search is client-side.
Frequently Asked Questions about LLM Wiki
What is LLM Wiki?
LLM Wiki is a Model Context Protocol (MCP) server that llm-powered knowledge base from your claude code, codex cli, copilot, cursor & gemini sessions. karpathy's llm wiki pattern — implemented and shipped. It connects AI assistants to external tools and data sources through a standardized interface.
How do I install LLM Wiki?
Follow the installation instructions on the LLM Wiki GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.
Which AI clients work with LLM Wiki?
LLM Wiki works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.
Is LLM Wiki free to use?
Yes, LLM Wiki is open source and available under the MIT license. You can use it freely in both personal and commercial projects.
LLM Wiki Alternatives — Similar Knowledge & Memory Servers
Looking for alternatives to LLM Wiki? Here are other popular knowledge & memory servers you can use with Claude, Cursor, and VS Code.
MemPalace
★ 52.6kA local AI memory system that stores all conversations verbatim and organizes them into navigable structures. It provides 19 MCP tools for AI assistants to search and retrieve past decisions, debugging sessions, and architecture debates automatically
Kratos
★ 25.7k🏛️ Memory System for AI Coding Tools - Never explain your codebase again. MCP server with perfect project isolation, 95.8% context accuracy, and the Four Pillars Framework.
Context Mode
★ 15.4kAn MCP server that preserves LLM context by intercepting large data outputs and returning only concise summaries or relevant sections. It enables efficient sandboxed code execution, file processing, and documentation indexing across multiple programm
Memu
★ 13.7kMemory for 24/7 proactive agents like OpenClaw.
MemOS
★ 9.3kMemOS (Memory Operating System) is a memory management operating system designed for AI applications. Its goal is: to enable your AI system to have long-term memory like a human, not only remembering what users have said but also actively invoking, u
Everos
★ 5.4kBuild, evaluate, and integrate long-term memory for self-evolving agents.
Browse More Knowledge & Memory MCP Servers
Explore all knowledge & memory servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.
Set Up LLM Wiki in Your Editor
Choose your AI client for step-by-step setup instructions.
Quick Config Preview
Add this to your claude_desktop_config.json or .cursor/mcp.json
Ready to use LLM Wiki?
Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.