Configuration

All settings can be changed through the web interface at /settings. You don't need to edit files manually.

The settings are saved in data/.env inside your mounted data directory.

Connection

Setting	What it does
Paperless-ngx URL	The address of your Paperless-ngx server
Paperless-ngx API Token	Found in Paperless-ngx under Settings → API Tokens

Choose the AI that reads and classifies your documents.

Enter your API key and choose a model (e.g. gpt-4o, gpt-4). Requires an account at platform.openai.com.

Enter the URL of your Ollama instance and the model name (e.g. mistral, llama3). Everything runs locally, no data leaves your network.

Requires your Azure endpoint, deployment name, API key, and API version.

For DeepSeek, OpenRouter, Perplexity, Gemini, LiteLLM, and others: enter the base URL, model name, and API key.

Setting	What it does
Scan interval	How often to check for new documents (default: every 5 minutes)
Assign tags	Whether the AI assigns tags
Assign correspondent	Whether the AI assigns a correspondent
Restrict to existing tags	AI only uses tags that already exist in Paperless-ngx
Restrict to existing correspondents	AI only uses correspondents that already exist
Process only tagged documents	If enabled, only documents with a specific tag are processed

Setting	What it does
Enable OCR Queue	Activates the OCR rescue feature for poorly scanned documents
Mistral API Key	Your Mistral AI API key

See OCR Queue for details.

These settings are configured via environment variables in your docker-compose.yml:

Variable	Default	Description
`TAG_CACHE_TTL_SECONDS`	`300`	How long to cache the tag list from Paperless-ngx (seconds)
`GLOBAL_RATE_LIMIT_MAX`	`120`	Max requests per 15-minute window per user
`GLOBAL_RATE_LIMIT_WINDOW_MS`	`900000`	Rate limit window in milliseconds
`API_KEY`	(empty)	Static API key for external integrations