Arthur MCPArthur MCP
Dashboard
How it worksFeaturesDocsEarly accessGitHubDashboard

How to Test a Tool

Verify a tool works before handing it to an AI client — much faster to debug from inside Arthur than from inside a chat window.

Prerequisites: a server with at least one tool.

Steps

  1. Open the server and go to its Tools tab.

    Open the capture server to document Test a Tool

  2. Select the tool you want to test — this opens its editor.

  3. Fill in the test arguments for the tool's parameters.

    Open the Tools tab, which contains the interface used by this procedure

  4. Run the test.

  5. Read the returned status and response body.

    Bring the New tool panel into the screenshot viewport

Confirm it worked

The test returns the status and body you expect for the given arguments. If it fails, the error message points at what went wrong (bad argument, upstream auth, unreachable endpoint).

Good to know

  • Testing calls the real upstream API — use test data or a sandbox credential if the action isn't safe to run for real.
  • The AI View tab on the server shows the server exactly as a connected AI client will see it, which is a good final check after testing individual tools.

Related

Tutorial video