Found!

Found! is a command-line tool for fast semantic search over your personal documents using vector embeddings and FAISS. It is designed to be simple, fast, and privacy-friendly, running entirely on your machine.

Features

• Semantic search for documents using natural language queries
• Fast similarity search powered by FAISS
• Caches document embeddings and index for speed
• Rich CLI interface with Typer

Installation and Usage with uv/uvx

Found! is compatible with uv and uvx for fast Python package management and execution.

1. Install uv (if not available)

See the official uv installation instructions for the recommended standalone installer and platform-specific details.

2. Install found with uv

uv tool install --git https://github.com/clssn/found.git

Usage

Search for a document

found doc "your query"

Options

• --document-dir, -d: Specify the directory containing documents (default: ~/Documents)
• --verbose, -v: Enable debug logging

Example

found doc "Tax certificate 2024" -d ~/Documents

How it works

• Recursively lists files in the specified document directory
• Generates semantic embeddings for each document using SentenceTransformers
• Builds a FAISS index for fast similarity search
• Caches the index and document list for future queries
• Returns the best matching document for your query

Requirements

• Python 3.8+
• faiss
• sentence-transformers
• typer
• platformdirs

License

MIT

Contributing

Pull requests and issues are welcome!

Made with ❤️ by clssn