Skip to main content
A deployment takes the code in a session and publishes it to a live URL. The deploy pipeline validates the project, builds a container, and launches it on infrastructure with HTTPS, health checks, and log streaming.

​
The deploy pipeline

Each step is available as a separate API endpoint so you can run them independently or chain them together:

​
What gets deployed

The deploy system examines the session’s workspace and detects the project type. It supports:
  • Node.js - Express, Next.js, React, and other Node frameworks
  • Python - FastAPI, Flask, Django, and other Python frameworks
  • Ruby - Rails applications
  • Static sites - HTML/CSS/JS served via a lightweight server
  • Docker - bring your own Dockerfile for any stack
Each project type has a handler that generates the right build steps, Dockerfile, and configuration.

​
Deploy manifest

Every deployable project needs a runtm.yaml at the workspace root:
If the session does not have one, call the scaffold endpoint and the system will generate one based on the detected project structure.

​
Streaming build logs

The POST /api/sessions/{id}/deploy/run endpoint streams NDJSON events as the build progresses:
This lets you show real-time progress in your UI or log the output for debugging.

​
Volumes

Sessions that need persistent storage across redeploys can configure volumes:
  • GET /api/sessions/{id}/deploy/volumes - read current volume configuration
  • PUT /api/sessions/{id}/deploy/volumes - update volume mounts
Volumes survive redeploys and are useful for databases, uploaded files, and caches.

​
Redeploying

Deploying again to the same project name updates the existing deployment. The system detects changes, rebuilds only what is necessary, and swaps to the new version with minimal downtime.

​
Permissions

Deploying requires the deployments:write scope on your API key. The preflight step also checks that the session’s owner has sufficient quota for the target tier. See Deploy Info for the full API reference.