Sends a question to a deployed agent and returns its answer. The response includes the output text, an interaction_id, and metadata with cited sources and any guardrail violations.

When to use it

Use it to call an agent from your own application. When the agent has authentication enabled, pass its API token in the x-api-token header.

Required parameters

• id (path): the application worker ID of the agent.

Response

A successful call returns 200 OK with output, the answer text; interaction_id; and metadata, which lists cited sources and guardrail violations and says whether a guardrail was violated.

Related: Ask an agent and stream the answer, Send feedback on an agent answer, Regenerate an agent API token