setToolResponse | SignalWire

Set the structured form of the response, separating what the tool did from what the model should do next. It replaces any string response, and the response becomes an object with up to two keys:

{
  "tool_result": "status: on hold",
  "tool_prompt": "Tell the caller you are placing them on hold."
}

Splitting them keeps the model from reading a status line aloud, and keeps the instruction from being mistaken for data. For a single string, use setResponse().

You can also set the structured form when you create the result: new FunctionResult({ tool_result: '...', tool_prompt: '...' }).

Parameters

toolResult

string

What the tool did or found: facts for the model to reason from, such as "payment declined" or "3 seats left". Omit it when there is nothing to report beyond the instruction.

toolPrompt

string

What the model should do next, such as "Tell the caller how many seats are left.". Omit it for a facts-only result.

Returns

FunctionResult — this, for chaining.

Example

import { AgentBase, FunctionResult } from '@signalwire/sdk';
const agent = new AgentBase({ name: 'my-agent', route: '/agent' });
agent.setPromptText('You are a helpful assistant.');
agent.defineTool({
  name: 'check_seats',
  description: 'Check remaining seats for a showing',
  handler: () => {
    const seats = 3; // look this up in your booking system
    return new FunctionResult().setToolResponse(
      `seats remaining: ${seats}`,
      'Tell the caller how many seats are left and ask how many they want.',
    );
  },
});
agent.run();