para-files Documentation
Welcome to para-files! This documentation is organized by how you want to use it.
I want to
Get Started
- Installation - Install para-files and verify it works
- Quick Setup - Configure in 5 minutes
- Classify My First File - Run your first classification
Use para-files
- CLI Overview - What commands are available
- Task Guides - How-to guides for common tasks:
Configure para-files
- Configuration Overview - How configuration works
- Environment Variables - Set via .env file
- YAML Config - Configure in personal_file_tree.yaml
Understand How It Works
- Architecture Overview - The big picture
- 6-Signal Pipeline - How files are classified
- Reference Tree - YAML structure explained
- MLX Embeddings - Semantic matching details
Troubleshoot Issues
- Files going to Inbox - Wrong classifications
- Low confidence scores - Adjusting thresholds
- Model download issues - Caching & performance
Develop & Extend
- Developer Guide - Building & contributing (in CLAUDE.md)
- Python API - Use para-files in your code
Quick Commands Reference
# Classify
uv run para-files classify document.pdf
# Move to PARA folders
uv run para-files move *.pdf --dry-run
# Add a bank
uv run para-files add-issuer "My Bank" -c banques
# Interactive learning
uv run para-files learn document.pdf
Key Features
- 6-Signal Pipeline: Validated DB → Rules → Book Detection → Domain KB → Semantic Router → LLM
- MLX Embeddings: Fast local semantic matching on Apple Neural Engine
- PARA Method: Organize into Projects, Areas, Resources, Archives
- Extensible: Add custom routes, issuers, and utterances via YAML
- Photo/Video Geotagging: GPS-aware organization by location
- Book Detection: Technical book identification via ISBN/metadata
System Requirements
- macOS with Apple Silicon (M1/M2/M3/M4)
- Python 3.12+
- uv package manager