Skip to main content
Langflow is a visual, low-code editor for building multi-agent and retrieval workflows in Python. This guide runs Langflow with its own Postgres database as one Levelrail app, with each service’s data in a named volume.

​
Deploy it

Then add a domain for the editor so it is served over HTTPS. See domains and ingress.

​
Security and requirements

Level: Standard. The template has no privileged-container finding. What this template needs:
  • Publishes ports directly on the host (langflow): 7860:7860.
  • Passwords and keys are generated at deploy time (db, langflow): LANGFLOW_DATABASE_URL, LANGFLOW_SUPERUSER_PASSWORD, POSTGRES_PASSWORD.
  • Stores data in named volumes (db, langflow): langflow_db_data, langflow_data.
The template sets LANGFLOW_AUTO_LOGIN to false, so the editor asks for the superuser login. The superuser password is generated, so you need to read it from the app’s secrets before your first sign-in.

​
What Levelrail adds

​
What a raw docker compose misses

  • The $SERVICE_PASSWORD_* placeholders in the template are resolved by Levelrail at deploy time and stored as secrets. A plain docker compose up cannot resolve them, so you would write the passwords yourself.
  • 7860:7860 is published on the host. The Langflow editor is reachable from the internet unless you filter it.
  • Compose does not back up either volume, so the flows and the Postgres data sit on one disk.
  • Compose has no rollback to an earlier revision.

​
Next steps

Last modified on October 20, 2026