Deploy a Windsurf app to production
Windsurf's Cascade agent edits, runs commands, and ships features without you touching the keyboard. Pair it with Launchmatic and `lm deploy` becomes the only command Cascade needs to know to put a feature in front of users.
Why Launchmatic for Windsurf
Cascade can run shell commands. Tell it the deploy command is `lm deploy` and the loop closes — describe a feature, Cascade implements it, Cascade ships it.
Deploy in 5 steps
- 1Open Windsurf and let Cascade scaffold or modify a project.
- 2Push to GitHub from Windsurf's source control panel.
- 3
npm i -g @launchmatic/cli && lm login. - 4Bind the repo:
lm init(single service) orlm monorepo init(workspace). - 5Add a Cascade rule that mentions
lm deploy(see snippet) so the agent uses it. - 6Trigger a Cascade run that ends with
lm deploy(orlm upfor the whole monorepo) and watch the URL appear.
Add to `.windsurfrules` so Cascade knows how to ship:
# Deployment
- The deploy command is `lm deploy`.
- Production logs: `lm logs --service web --follow`.
- Roll back a bad deploy: `lm deployments list`, then `lm deployments rollback <deploymentId>`.FAQ
Can Cascade roll back a bad deploy on its own?+
Yes — `lm deployments rollback <deploymentId>` reverts to a previous successful deploy (find the id with `lm deployments list`). Add both to your `.windsurfrules` and Cascade can self-heal failed deploys when test suites fail post-ship.
Ready to deploy?
Free to start. No credit card. Auto-SSL on custom domains, managed Postgres, and per-branch preview deploys included.
This guide is part of our complete vibe coding hosting guide — how to take any AI-built app to production.