Homelib
Your knowledge, your library, your machine.
Your bookmarks, starred repos, and saved articles live on someone else's servers. The content behind them disappears behind paywalls, gets edited, or vanishes. Homelib gives you a private knowledge library on your own machine. Save a URL and it reads the content for you — fetches it, summarizes it, extracts entities, and makes everything searchable.
The Problem
Your notebook is for your thoughts — your projects, your original ideas. But the things you discover online? Articles, tutorials, repos, references? They don't belong there. And bookmarks don't work either: links break, pages get paywalled, and even when they work, you have to click through, navigate the page, dismiss the cookie banner, and re-read the whole thing to find what you were looking for.
Think about how it worked 200 years ago. You had your diary — your personal journal. And you had your books — a reference shelf of sources you collected for research and learning. Two separate things, two different purposes.
Homelib is the reference shelf for your digital life.
How It Works
Save a URL and Homelib reads it for you. It fetches the content, converts it to clean Markdown, and analyzes it with AI — classifying, tagging, extracting people, companies, products, and places. Everything gets indexed and searchable.
Import & Catalog
Bring in bookmarks from Safari Reading List, GitHub Stars, and more. Content is automatically fetched, cleaned, analyzed, and made searchable.
Go Deeper
Start with a bookmark of a DIY project. Homelib discovers the products used, the tools, the people behind it. It researches these through web searches and builds knowledge on top of that item.
Knowledge Compounds
Those discoveries become new items in your library. As it grows, structure emerges: a directory of people, products, companies — all linked and searchable.
Explore What Emerges
As your library grows, you don't just have bookmarks — you have entities, topics, and connections that span across everything you've saved.
Built for You
Fully Local
Everything runs on your machine. All data stored in plain files — YAML, Markdown, JSON. Human-readable, git-friendly, portable.
Search Everything
Full-text search with field filters: tag, type, domain, author, company, product. Find anything in your collection instantly.
AI Chat Agent
Ask questions and get answers grounded in your own collection. The agent searches, compares items, and summarizes topics.
MCP Integration
Connect to Claude Desktop, Cursor, or any MCP-compatible tool. Add items and query your library from any chat interface.
Integrates With
Open Source
Homelib runs as a single Python server — no separate frontend and backend, no job queues, no cloud dependency. A lightweight UI with Alpine.js and htmx. The database rebuilds from the files. The files work with any tool.