Running tests
Running tests individually
Open a test and click Run. You will choose:
Branch - the Bubble version you want to test.
Model - the model Buildprint should use for the run (see Best practices for guidance on selecting a model)
If a test user is assigned for the selected environment, Buildprint uses that account for the run. If no user is assigned, the run still starts, but the agent will not have a pre-configured login account.
Live runs use the Live version user; non-live branch runs use the Test version user.
Running test groups
Run a group from the group menu in the sidebar or from the group page. You choose the branch and model once, then Buildprint starts one child run per test in that group.
Group runs are useful for broader regression checks because all member tests are launched in parallel and tracked together under one parent run.
Selecting a model
Use a model with vision, since the agent is working against a live browser. In general, start with the cheapest model that is capable of reliably passing your tests.
In practice, Sonnet 4.6, GPT 5.4, and Gemini 3 Flash tend to be good enough for most test suites.
Running tests is currently done from the Buildprint UI. API support for test execution is planned for the future.