# Test an Action

> Run one of an Action's operations from the editor and see what the API sends back, before visitors use it.

Source: https://www.chirpbuilder.com/help/actions/test  
Plans: Essentials, Creator, Business, Academy  
Last checked: 2026-10-27

When you're done, you know the Action works and what the AI gets back from it.

**Who can do this:** the account owner on any plan, and workspace admins and editors.

## Run a test

1. Open the Action in **7. Actions** and click **Test** under an operation.
2. Fill in its values. A * means required; the box shows the allowed values or the type. If it sends a
   body, it's filled in with the fields as a start.
3. For an operation that reads data, click **Run test** (1). It calls the real API.

![The test panel of findForecastArea with the point 45.52,-122.68 typed in, Run test (1), and the result: Worked (HTTP 200) with the forecast office and grid position weather.gov returned.](https://mjmtbasmrtksluvlnema.supabase.co/storage/v1/object/public/help-media/web-optimized/actions/test/SS-ACT-050-01__test-result__essentials__desktop__v01.webp?v=a09b8d3066)

The result shows the request (a key in the web address shows as [hidden]), whether it **Worked** or why
not, the HTTP status, how long it took, and what the API sent back. The AI gets that response too, up to 20,000
characters; the panel shows the first 6,000.

## Testing a change

For an operation that changes data:

- **Preview request** shows what would be sent, with its headers (keys as [hidden]), without sending
  anything. A long body is shortened.
- **Send for real** sends it, after your browser asks "Send a real … request to …? It may create, change
  or delete data there."

## Good to know

- **Tests use your latest, unpublished settings**, and work even when the operation or Actions are
  switched off.
- **Tests cost nothing on OpenRouter**, since no AI model is used. The API's provider may charge.
- **Real test calls show in Recent calls** as Test: see
  [Recent calls and failing-call alerts](https://www.chirpbuilder.com/help/actions/recent-calls). A preview that's fine isn't logged; one with bad
  values, or one that's blocked, shows as Test.
- **Last test failed** shows when the latest test of an operation that's switched on failed, bad values
  included. It goes once a new test of that same operation works, or you switch that operation off, or
  the schema is replaced or read again. **Needs an API key** shows instead while a key is missing.
- **You can run up to 60 tests in each clock hour**, previews included. Tests your AI app runs count
  too.

## If it doesn't work

- **API error (HTTP 401 or 403):** the key is wrong or missing: see
  [Add the API key an Action needs](https://www.chirpbuilder.com/help/actions/api-key).
- **API error (HTTP 400 or 404):** check the values you typed.
- **Bad values:** a value didn't match what the schema expects.
- **Blocked:** the address isn't allowed (it must be a public https address), or it redirected to
  another address.
- **"The Action "…" needs an API key. Add it in the Chirp builder (Actions) first."**: see
  [Add the API key an Action needs](https://www.chirpbuilder.com/help/actions/api-key).
- **Timed out:** the API didn't answer within 20 seconds.
- **Couldn't connect:** the API's address couldn't be reached, or it redirected too many times.
- **"The request body isn't valid JSON."**
- **"That's a lot of tests. Please wait a few minutes and try again."**