RAG-optimized personal knowledge management system.
VaultMind is a desktop app built with Tauri 2.0 where Retrieval-Augmented Generation (RAG) is a first-class architectural concern — not a plugin afterthought. Every note you create is automatically parsed, chunked, embedded, and connected in a knowledge graph for high-quality AI-powered retrieval.
| Obsidian + Plugins | VaultMind | |
|---|---|---|
| Chunking | Naive (per-file/heading) | AST-based with hierarchical context |
| Embeddings | Plugin dependency | Built-in (ONNX, local-first) |
| Knowledge Graph | Link graph only | Explicit (wikilinks) + implicit (semantic similarity) |
| AI Agent | Chat-only | Action-capable (create, edit, link, refactor notes) |
| Retrieval | Vector search only | Hybrid (vector + BM25 + graph expansion + re-ranking) |
| Architecture | Plugin bolt-on | RAG as core design principle |
| Layer | Technology |
|---|---|
| App Shell | Tauri 2.0 (Rust) |
| Frontend | Svelte 5 + TipTap editor |
| Vector DB | LanceDB (embedded) |
| Metadata | SQLite (rusqlite) |
| Full-Text | Tantivy |
| Embeddings | ONNX Runtime (all-MiniLM-L6-v2) |
| Graph | petgraph |
# Clone the repository
git clone https://github.com/YOUR_USERNAME/vaultmind.git
cd vaultmind
# Run the setup script (installs dependencies, hooks, etc.)
./scripts/setup-dev.sh
# Start development
pnpm tauri dev
# Build for your platform
pnpm tauri build
vaultmind/
├── .github/ # CI/CD, issue templates, PR template
├── src-tauri/ # Rust backend
│ ├── Cargo.toml # Workspace root
│ ├── crates/
│ │ ├── core-storage/ # File I/O, SQLite, LanceDB
│ │ ├── indexer/ # AST parser, chunker, embedder
│ │ ├── graph-engine/ # Knowledge graph
│ │ ├── retriever/ # Hybrid search, re-ranking
│ │ └── agent-runtime/ # LLM integration, tool execution
│ └── src/main.rs # Tauri entry point
├── src/ # Svelte frontend
│ ├── lib/components/ # Reusable UI components
│ ├── lib/stores/ # State management
│ └── routes/ # App views
├── models/ # ONNX models (git-lfs)
├── docs/ # Architecture docs, ADRs
├── tests/ # Integration & E2E tests
└── scripts/ # Dev scripts, benchmarks
We welcome contributions from both humans and AI agents. See CONTRIBUTING.md for guidelines.
| Branch | Purpose |
|---|---|
main |
Stable releases |
develop |
Integration branch |
feature/* |
New features |
fix/* |
Bug fixes |
MIT — see LICENSE for details.