Skip to content
DocsOpen Hub
Practical guides

Take an API or app live

On this page

Treat launch as a review of behavior, configuration, and recovery. A successful sandbox request is the starting point; the production client and resources still need verification.

Confirm the API contract#

Review the public hostname, controller path, action route, method, required inputs, response shape, and failure behavior. Keep existing clients compatible or plan the transition to a changed contract.

Test success, invalid input, empty results, denied access, and a failed dependency. For an app, test the actual page interaction as well as its underlying request.

Configure production resources#

  • Set production selector values and database bindings.
  • Confirm gateway credentials, action scope, and request limits.
  • Verify IdP authorization and record ownership checks for private data.
  • Configure the app's exact browser origins under CORS Origin.
  • Review country/IP controls independently.
  • Review schedules, notification recipients, and external service modes.

Do not assume a sandbox prefix isolates every downstream resource. Identity, email, and external services can be shared.

Review the release#

Use Workflows → Releases for workflow candidates and Web Apps → Releases for app candidates. Inspect differences, source information, and dependencies. Record the purpose of the release and the tests that support it.

For a database change, confirm the backup and compatibility plan. Know which prior workflow/app release you can restore and what data recovery would still be needed.

Verify the live path#

After publication, make a controlled request through the real production client. Confirm the returned records or result, inspect the corresponding log, and check the app's displayed state.

For external writes, verify the provider's recorded operation rather than treating a gateway acknowledgement as final completion.

Monitor and hand over#

Watch errors, latency, usage, and scheduled work over the relevant interval. Record the released version, owner, known limitations, and escalation path. If something regresses, use the evidence to decide whether to fix forward or restore a known release.

Need to investigate a problem? Start with troubleshooting