Get the full Sovereign Workflows stack running on your infrastructure in under 10 minutes using Docker Compose.
.lic file) — obtain one from your Portal dashboard4723 available (reverse proxy — the only port you need to access)Download the Docker Compose configuration and environment template:
Or use curl:
curl -O https://portal.nemorion.com/downloads/docker-compose.yml
curl -O https://portal.nemorion.com/downloads/.env.exampleinit-db.sql creates the Auth database on first Postgres start. Place all three files in the same directory.Copy the example file and set the required values:
cp .env.example .env| Variable | Description | How to Generate |
|---|---|---|
POSTGRES_PASSWORD | Database password | openssl rand -base64 32 |
RABBITMQ_PASSWORD | Message queue password | openssl rand -base64 32 |
CONNECTOR_GLOBAL_SERVICE_KEY | Executor-to-connector auth | openssl rand -base64 32 |
Everything else has sensible defaults. LLM Chat works out of the box with a Nemorion-provided API key — no configuration needed.
Copy your .lic file to the same directory as docker-compose.yml:
# Default expected path (configurable via LICENSE_FILE_PATH)
cp ~/Downloads/nemorion-workflows-enterprise.lic ./license.licdocker compose up -dDocker will pull images from GHCR, start infrastructure services first (Postgres, RabbitMQ, Redis), then application services once health checks pass. First run takes 2-3 minutes for image pulls.
Monitor startup progress:
docker compose ps
docker compose logs -fexecutor-worker starts 2 replicas by default. Adjust with EXECUTOR_REPLICAS=4 in your .env for higher throughput.Once all containers are running, open your browser to:
http://localhost:4723You should see the Workflows UI. Click "New Workflow" and verify that connector actions are displayed — this confirms the connectors registered successfully.
You can also verify individual services directly:
docker compose ps # All containers should be "healthy"
curl http://localhost:4723 # Reverse proxy → UIRabbitMQ management UI is at http://localhost:15672 (login with your configured credentials).
The visual editor is available at http://localhost:4723. From here you can:
Connect your AI agent (Claude, GPT, etc.) to the MCP Server via the reverse proxy at http://localhost:4723/mcp/. The server exposes all workflow actions as tools that AI agents can invoke directly.
If a port is already in use, change it in your .env:
# Example: move Postgres to port 5433
POSTGRES_PORT=5433Check the logs for the specific service:
docker compose logs engine-api
docker compose logs auth-apiIf the Engine API logs show license validation failures:
license.lic exists in the expected pathdocker compose down -v