Libro - The open-source brain for AI agents.

👋 Hey Product Hunt! I'm Lalit, the creator of Libro. I love building AI tools, but I was incredibly frustrated by one thing: AI agents have goldfish memory. Every time I started a new chat session with Claude or Cursor, I had to repeat the exact same context. "Use Tailwind," "Here is my database schema," "I prefer this coding style." I realized we needed a dedicated User Context Layer—an operating system for AI memory. So, I built Libro. Libro is an open-source, drop-in memory infrastructure layer for your AI applications. Instead of shoving massive system prompts into every API call, Libro sits quietly in the background: 🧠 Zero-cost Local Embeddings: We handle vectorization locally so you don't pay OpenAI per token. 🧩 Semantic Deduplication: As user preferences evolve, Libro automatically prunes outdated context. 🔌 Universal MCP Server: Plug Libro directly into Cursor, Windsurf, or Claude Desktop so your local AI instantly knows everything about you. 🌐 Chrome Extension: Extract context "passports" from any website or codebase with one click. I built this because I wanted AI that actually remembers me. I've open-sourced the entire platform because I believe memory should belong to the user, not locked inside an LLM's black box. I would absolutely love for you to try out the SDK, spin up the MCP server in your local editor, and let me know what you think! I'll be hanging out in the comments all day to answer any questions. Cheers, Lalit