@harshitpadha/kb-wiki
v1.0.1
Published
Next-generation implementation of Karpathy's LLM Wiki with MCP and local semantic search.
Maintainers
Readme
🧠 Knowledge Base Wiki
A next-generation implementation of the LLM Wiki concept proposed by Andrej Karpathy, designed to turn your local Markdown files into a dynamic, semantically searchable knowledge base maintained by an AI.
🚀 Features
- 📁 Local-First Markdown: Your knowledge base lives as plain Markdown files in your filesystem. No lock-in, fully version-controllable.
- 🔍 Semantic Search: Powered by
sqlite-vec, files are chunked and embedded in a local SQLite database for lightning-fast semantic queries. - 🤖 MCP Server Integration: Exposes your knowledge base directly to MCP-compatible AI assistants (like Claude Desktop or Cursor).
- 👷 Dynamic YAML Subagents: Define specific roles (e.g.,
Librarian,Archivist) using simple.yamlfiles. - 🛠️ Auto-Registration: The CLI automatically connects the MCP server to Claude Desktop, Gemini CLI, and Claude Code.
📦 Quickstart
1. Installation
You can install the Knowledge Base Wiki globally using npm:
npm install -g @harshitpadha/kb-wikiOr run it directly using npx:
npx @harshitpadha/kb-wiki init2. Initialize a Knowledge Base
Navigate to any directory where you want to manage your Markdown vault and run:
kb init3. Start Chatting
Open Claude Desktop, Gemini CLI, or Claude Code. The /kb slash commands will be ready to use!
📚 Documentation
Check out the full wiki for in-depth guides:
📜 License
This project is licensed under the MIT License - see the LICENSE file for details.
