v0.3.2 — A high-performance desktop application for forensic document analysis, built with Tauri 2 + Rust + SvelteKit 5.
All processing runs locally — no cloud dependencies.
- Ingest PDFs, images, audio, DOCX, and plain-text evidence files
- Local LLM inference (llama.cpp, GGUF models, Metal GPU on Apple Silicon)
- Multi-pass analysis pipelines (Basic Facts, Financial Crimes, Document Analysis, OCR, Audio)
- Document metadata extraction: EXIF, PDF properties, language detection, structured fields
- Fact deduplication, quality scoring, annotation, and tagging
- Timeline, network graph, geographic map, and statistics visualizations
- Export to JSON, CSV, PDF, and Excel
- Backup/restore with optional evidence file bundling
See docs/README.md for comprehensive documentation.
- Getting Started
- System Architecture
- Tauri Commands API
- Database Schema
- Release Process
- Contributing
- AGENTS.md — build/test commands and architecture notes for contributors
# Install JS dependencies
npm install
# Start full Tauri dev build (Rust + frontend hot-reload)
npm run tauri dev
# Type-check frontend
npm run check
# Type-check backend
cd src-tauri && cargo checkmise users:
mise run setup && mise run devalso works if you have mise installed.
AGPL-3.0-only (see LICENSE file)