Build the dataset easily for RAG evaluation & optimization in AutoRAG family.

Desktop application for building and managing RAG evaluation datasets, works perfect with AutoRAG-Research Ingest documents, annotate queries with evidence, and export structured datasets. Currently supports Visual Document Retrieval type datasets, more in the future.

• PDF/image ingestion with page-level rendering and thumbnail caching
• Query annotation — query text, query-to-LLM prompts, generation ground truths
• Evidence grouping with graded relevance scoring
• Export to CSV, JSON, or ZIP bundles

• Tauri v2 prerequisites (Rust, system deps)
• pnpm
• Docker (for PostgreSQL)

# Start the database
make docker-up

# Install dependencies and run
pnpm i
pnpm tauri dev

Tauri v2 (Rust) · Vue 3 + TypeScript · Vite · Tailwind CSS v4 · Pinia · shadcn-vue · pnpm

src/              Vue frontend
src-tauri/        Rust backend (Tauri commands, DB access)
postgresql/       Docker Compose + init scripts for PostgreSQL + VectorChord

pnpm tauri dev        # Dev mode (frontend + backend)
pnpm tauri build      # Production build
pnpm test             # Run Vitest tests
pnpm type-check       # TypeScript checking
pnpm exec eslint .    # Lint frontend
cargo fmt             # Format Rust (in src-tauri/)
cargo clippy          # Lint Rust (in src-tauri/)