Skip to content

Testing

Testing helps confirm that your project changes behave as expected before they are pushed, reviewed, or merged.

There is no local runtime

The ADK does not execute your agent on your local machine. There is no poly serve command or local simulator. All agent execution happens inside Agent Studio's sandbox environment. To test runtime behavior, push your changes and use poly chat, or test interactively through Agent Studio.

In the ADK workflow, testing usually sits alongside validation and manual review in Agent Studio.

What testing is for

Testing is useful when you want to:

  • check that code changes behave as expected
  • catch regressions before pushing
  • validate function logic outside the runtime conversation loop
  • support safer review and collaboration

  • Validation

    Use poly validate to check project configuration before pushing.

  • Runtime review

    Use Agent Studio and poly chat to test behavior interactively.

Testing in the workflow

A typical development loop looks like this:

  1. edit files locally
  2. inspect changes with poly status and poly diff
  3. run poly validate
  4. push changes with poly push
  5. test the branch in Agent Studio

What to test

The exact tests will depend on the kind of work you are doing, but common areas include:

  • function logic
  • state transitions
  • API integration helpers
  • formatting or normalization utilities
  • project-specific edge cases

Best practices

  • use validation as part of the normal workflow, not just before merge
  • test important error paths, not only success cases
  • combine interactive review when behavior depends on conversation flow