Side-car for paperless-ngx
Tag a document.
Get it back filed.
It reads the documents you tag and writes back a clean title, tags, correspondent and date, using the OCR and language models you bring.
- Bring your own keys
- Self-hosted
- One container
- No telemetry
How it works
One tag in. A filed document out.
Tag a document
Add
psf-processin paperless. That one tag is the whole trigger.OCR
optionalA vision model re-reads the scan, or skip it and reuse paperless’s own text.
Extract
One model call returns the title, tags, correspondent and date.
Apply or review
Written back to paperless automatically, or queued for your approval.
Re-tag to run a document again. Unchanged content is skipped, so re-runs are free.
Screenshots
A look at the app.
Bring your own keys
Use your own API keys.
Connect the providers you already pay for, or a model running on your own hardware.
Encrypted at rest, and sent straight to the provider, never through us.
Features
Everything it does.
Better OCR
Re-read scans with a vision model when paperless’s own text falls short.
Automatic metadata
A clean title, tags, correspondent and date for every document.
Review or auto-apply
Approve, edit or reject each suggestion, or write it straight back.
Prompts you control
Edit what each model is asked, and test it on a real document first.
Any provider
Anthropic, OpenAI, Google, Mistral, or any local OpenAI-compatible model.
Full history
See the exact prompt sent and the model’s reply behind every run.
Easy to set up. Easy to use.
Free, open source and self-hosted. Connect your keys, tag a document, and let it do the filing for you.