Run the editor locally

Goal: start the editor and its Python backend with Docker Compose. Use Docker with Compose v2 and Git. The hosted editor does not need this setup.

Clone and configure

git clone --recurse-submodules https://github.com/BESSER-PEARL/BESSER.git
cd BESSER

Copy the root .env.example to .env and set values for your local deployment. The frontend service also reads besser/utilities/web_modeling_editor/frontend/packages/webapp/webpack/.env; the legacy compose definition expects this file even though its directory is not shipped. Create the directory and file with the local URLs below.

DEPLOYMENT_URL=http://localhost:8080
BACKEND_URL=http://localhost:9000/besser_api
UML_BOT_WS_URL=ws://localhost:8765

Frontend URLs are baked into the JavaScript at build time. The compose env_file alone cannot change them. Supply these values in besser/utilities/web_modeling_editor/frontend/packages/webapp/.env before building, or pass them as Docker build arguments. See Environment Variables for the complete variable reference.

Build and start

docker compose up --build -d
docker compose ps

Open http://localhost:8080. Check http://localhost:9000/besser_api/ for a backend response, then create a class diagram and try Quality Check.

The root compose file starts the frontend, backend, and agent simulator. Its modeling-agent service is commented out: the conversational assistant needs that service started separately. Follow the Modeling Agent setup guide to configure it before using AI modeling.

Check logs and stop

docker compose logs --tail=100 besser-wme-backend
docker compose logs --tail=100 besser-wme-frontend
docker compose down

For source development, use Run the editor from source. For a hosted v8 deployment with the separate generation worker, use the production runbook.