Testing & Debugging Actions

End-to-End Testing in the Playground

The Playground lets you test your Action directly inside Watermelon, without publishing it. It simulates how your AI Agent will call your API.

How to Test

Go to AI Agent → Playground

Start having your converstion

See results

When an action is called this is shown in the Playground as source under the agent response

Comparing a Working Postman Call to a Failing Action

When your Action fails but works in Postman, something in Watermelon’s setup doesn’t match your Postman request.

Step-by-step comparison

Element	In Postman	In Watermelon	What to Check
Base URL	Matches documentation?	Must be identical	e.g., https://api.example.com/v1
Auth	Token, header, or basic auth	Configured in Authentication section	Matching case & syntax
Path	Exact endpoint	Defined in schema under paths:	Same structure and case
Query / Body	Key-value pairs	Input fields	Correct names and types
Headers	Content-Type, Authorization, etc.	Only if required	Should not duplicate system headers

If your Postman call succeeds but the Action fails:

Compare full request URL and headers in both.

Ensure schema uses correct parameter location (in: query, in: path).

Check for case sensitivity — orderId ≠ order_id.

End-to-End Testing in the Playground