RAGFS
A FUSE filesystem for semantic search using vector embeddings
Features
- Semantic Search
- Local Embeddings
- FUSE Integration
- Real-time Indexing
- Multimodal Support
- Code-aware Chunking
- Hybrid Search
- Python Bindings
- Agent File Operations
- Atomic Transactions
Quick Start
# Index a directory
ragfs index ~/Documents
# Semantic search
ragfs query ~/Documents "authentication logic"
# Mount as filesystem
ragfs mount ~/Documents ~/mnt --foreground
# Check index status
ragfs status ~/DocumentsArchitecture
Crate Documentation
ragfs
CLI application with index, query, mount, and status commands
ragfs-core
Core traits and types: VectorStore, Embedder, Chunk
ragfs-extract
Content extraction from text, code, PDF, and images
ragfs-chunker
Fixed-size, semantic, and code-aware chunking strategies
ragfs-embed
Local embedding generation using Candle (gte-small, 384-dim)
ragfs-store
Vector storage with LanceDB and hybrid search
ragfs-index
Indexing pipeline and file watching
ragfs-query
Query parsing and execution
ragfs-fuse
FUSE filesystem implementation with virtual .ragfs/ directory
ragfs-python
Python bindings with PyO3 for LangChain, LlamaIndex, and Haystack