Docker ready

Your private document vault is about fifteen minutes away.

DocManFu packages the web app, API, database, task queue, and OCR worker into one Docker Compose stack.

Clone the repository

You need Git, Docker Engine, and Docker Compose v2.

git clone https://github.com/DocManFu/DocManFu.git
cd DocManFu

Create your configuration

Copy the production example and set a strong PostgreSQL password and JWT secret. The deployment guide documents every option.

cp .env.production.example .env
$EDITOR .env

Required: Change POSTGRES_PASSWORD and JWT_SECRET_KEY before starting.

Start DocManFu

./prod up -d
./prod ps

The first build downloads images and prepares OCR dependencies. Open http://localhost:8080 when the services are healthy.

Create your administrator

The first-run setup screen creates the admin account. Sign in, upload a document, and watch OCR progress in real time.

Optional: enable AI

DocManFu works without AI. Add OpenAI, Anthropic, or Ollama when you want automated names, types, tags, and metadata.

AI_PROVIDER=ollama
AI_BASE_URL=http://host.docker.internal:11434
AI_MODEL=llama3.2

Scanner automation

Install the folder watcher.

The optional docmanfu CLI watches a directory and uploads new PDFs or images. Point a network scanner at that folder and filing becomes pleasantly boring.

brew tap DocManFu/docmanfu
brew install docmanfu
docmanfu --path ~/scans

SCANNER→~/scans→

DocManFuWatches quietly · uploads automatically

Your documents. Your rules.

Ready to master the paper pile?

One Docker Compose command puts a private, searchable document vault on hardware you control.

Start self-hosting View on GitHub ↗