Ask the agent

Run the NeuroAPI agent and return the answer.

Defaults: mode=smart, stream=false. With stream=false the response is a single JSON document. With stream=true it is an SSE stream (text/event-stream) where each event is one TaskResponseMessage from the agent and the stream terminates with data: [DONE]\n\n on success.

Optional system_prompt augments the agent's instructions (it never overrides its persona or safety guardrails). Optional output_schema (a JSON Schema) makes the agent return a validated object in output (answer is then null - exactly one is set); it requires stream=false.

Every response carries an X-Request-Id header that mirrors AgentAskResponse.request_id. Pass it back when reporting issues.

curl -X POST "https://api.neurobro.ai/api/v1/agent/ask" \  -H "Content-Type: application/json" \  -d '{    "mode": "smart",    "prompt": "Summarise the latest macroeconomic outlook for the eurozone.",    "stream": false  }'

{
  "answer": "The European Central Bank held rates steady ...",
  "mode": "smart",
  "request_id": "8b3a2f0e-1d7e-4c6b-9b2a-f5e23a1c9d44",
  "usage": {
    "cached_tokens": 80,
    "completion_tokens": 256,
    "cost_units": 2,
    "prompt_tokens": 124
  }
}