Self-Hosted / Projects / Paperless-ngx
Paperless-ngx
StableSpecs verified August 25, 2026 · against v3.0.5
Overview
Self-hosted document management system for OCR, tagging, and searchable digital archives.
Paperless-ngx is a self-hosted document management system for people who want to turn paper mail, invoices, statements, and scans into a searchable digital archive they control.
Why people choose Paperless-ngx
The project is popular with homelab users because it combines OCR, metadata extraction, tags, correspondents, document types, and full-text search in one interface instead of making you sort PDFs manually in folders.
What is included
Official docs and the README highlight OCR with Tesseract, support for PDF and image imports, optional Tika-based Office document support, machine-learning suggestions for tags and document types, email ingestion, shareable links, custom fields, saved views, and bulk editing from the web app.
What to know before you deploy
Paperless-ngx works best on a trusted server with backups in place. The maintainers explicitly warn against running it on an untrusted host because documents are stored in clear text without encryption, and OCR or AI-assisted features can feel slow on low-power hardware like older Raspberry Pi systems.
Best fit
Choose Paperless-ngx when you want a long-term home archive for household paperwork or a small-team document hub with search and metadata. If you only need a synced file folder with no OCR pipeline, lighter file storage tools may be enough.
Interface previews
Screenshots of the Paperless-ngx interface.
Feature support
| Feature | Support |
|---|---|
| OCR for scanned documentsOfficial docs say Paperless-ngx performs OCR with Tesseract and can recognize more than 100 languages. | Supported |
| Full-text search and highlightingThe docs highlight full-text search with autocomplete, relevance sorting, and highlighted matches. | Supported |
| Tags, correspondents, and document typesMetadata organization is a core feature in the official docs and web UI. | Supported |
| Machine-learning suggestionsOfficial docs say Paperless-ngx can automatically suggest tags, correspondents, and document types. | Supported |
| Email ingestionThe docs describe importing documents from email accounts with post-processing actions. | Supported |
| Office document supportOffice files are supported when optional Tika integration is enabled. | Partial support |
| Multi-user permissions and workflowsOfficial docs list multi-user permissions and a workflow system. | Supported |
| Opt-in AI document assistanceThe docs describe LLM-powered suggestions and chat as optional features that are disabled by default. | Partial support |
Community signals
- Stars
- 46.5k
- Forks
- 3.2k
- Open issues
- 5
- Last commit
- 19h ago
- Latest release
- v3.3.0
- Repo created
- Feb 2022
Includes open pull requests
6d ago
Refreshed nightly from the GitHub API.
Questions
What is Paperless-ngx used for?
Paperless-ngx is used to scan, import, OCR, tag, and search personal or business documents from a self-hosted archive.
Does Paperless-ngx do OCR?
Yes. The official docs say Paperless-ngx performs OCR with Tesseract and adds searchable text even to image-only scans.
What file types does Paperless-ngx support?
Official docs say it supports PDF documents, common image formats, plain text files, and Office documents when optional Tika integration is enabled.
Does Paperless-ngx support Docker?
Yes. The setup docs recommend Docker or Docker Compose for most users and also provide an installation script for Docker-based setups.
Can Paperless-ngx run on Raspberry Pi?
Yes, but the FAQ warns that OCR can be slow on Raspberry Pi hardware, so stronger hardware is more comfortable for heavier document workloads.
Can you export your files later if you switch tools?
Yes. The FAQ explains that documents are stored as plain files on disk and can be exported, with configurable naming or the built-in exporter.
Is Paperless-ngx encrypted by default?
No. The maintainers explicitly warn that Paperless-ngx stores information in clear text without encryption and should only run on a trusted host.
support // the lab
Found this write-up useful?
If it saved you time or a rebuild, you can support more practical homelab guides.