Integration Overview

How to configure DocuStore's external dependencies: authentication, LLM providers, blob storage, and observability.

DocuStore integrates with several external services. Each integration follows the port-adapter pattern -- the application layer defines an abstract port (protocol), and the infrastructure layer provides one or more concrete adapters. This means you can swap providers without changing business logic.

Integration Points

Integration	Port	Adapters	Config Prefix
Authentication	`RequestAuth`	Sentinel SDK	`SENTINEL_*`
LLM Providers	`LLMProvider`	Ollama, OpenAI, Gemini	`LLM_`, `CHAT_LLM_`
Blob Storage	`BlobStore`	Local filesystem, S3	`BLOB_*`
Observability	`PromptRepository`	Langfuse, YAML fallback	`LANGFUSE_*`

Configuration

All configuration is managed through environment variables, read by a single Settings class (pydantic-settings). Variables can be set in a .env file at the project root or passed directly as environment variables.

# .env file example
APP_ENV=development
LOG_LEVEL=INFO

# Core infrastructure
EVENTSTOREDB_URI=esdb://localhost:2113?tls=false
MONGO_URI=mongodb://localhost:27017/?replicaSet=rs0
QDRANT_URL=http://localhost:6333
TEMPORAL_ADDRESS=localhost:7233
KAFKA_BOOTSTRAP_SERVERS=localhost:19092

# LLM
LLM_PROVIDER=ollama
LLM_MODEL_NAME=gemma3:27b
LLM_BASE_URL=http://localhost:11434

# Auth
SENTINEL_URL=http://localhost:9003
SENTINEL_SERVICE_KEY=your-service-key

# Observability
LANGFUSE_HOST=http://localhost:3000
LANGFUSE_PUBLIC_KEY=pk-lf-docu-store-dev
LANGFUSE_SECRET_KEY=sk-lf-docu-store-dev

Environment Modes

DocuStore supports three application environments:

Mode	`APP_ENV`	Typical Use
`development`	Default	Local Docker Compose stack
`staging`	Staging	Pre-production testing
`production`	Production	Live deployment

The environment mode affects default values and logging behavior but does not change the API surface. All features are available in all modes.

Quick Reference: Essential Variables

Variable	Default	Description
`APP_ENV`	`development`	Application environment
`LOG_LEVEL`	`INFO`	Logging level
`API_HOST`	`127.0.0.1`	API server bind address
`API_PORT`	`8000`	API server port
`EVENTSTOREDB_URI`	`esdb://localhost:2113?tls=false`	EventStoreDB connection
`MONGO_URI`	`mongodb://localhost:27017/?replicaSet=rs0`	MongoDB connection
`QDRANT_URL`	`http://localhost:6333`	Qdrant vector database
`TEMPORAL_ADDRESS`	`localhost:7233`	Temporal server
`KAFKA_BOOTSTRAP_SERVERS`	`localhost:19092`	Kafka brokers
`BLOB_BASE_URL`	`file://./blobs`	Blob storage root
`LLM_PROVIDER`	`ollama`	LLM backend
`SENTINEL_URL`	`http://localhost:9003`	Sentinel auth service

Docker Compose

For local development, all infrastructure services are provided via Docker Compose:

# Start all services (EventStoreDB, MongoDB, Qdrant, Temporal, Kafka, Langfuse)
make docker-up

# This runs both docker-compose.yml and docker-compose-langfuse.yml
# on the shared docu_store-network

The Makefile provides convenience targets for all common operations:

make run                # Start API server
make run-workflow-worker # Start pipeline worker
make run-temporal-worker # Start Temporal worker
make docker-up          # Start infrastructure
make docker-down        # Stop infrastructure