Ad-Hoc Tools Schema Resolver connector

The Ad-Hoc Tools Schema connector is an outbound connector that implements the tool resolution part of the AI Agent connector.

About this connector

Although its function is also embedded in the AI Agent connector, the Ad-Hoc Tools Schema connector can be used independently in combination with other AI connectors.

This can be useful for:

• Direct LLM interaction: If you don't want to use the AI Agent connector but still want to resolve tools for an ad-hoc sub-process, use the fromAi function to define the input schema in combination with custom LLM integrations.
• Debugging tool definitions: If you want to test the tool resolution logic without having to set up a full AI Agent connector, you can use this connector to get the tool definitions and see how they are generated.

Prerequisites

The following prerequisites are required to use this connector:

Prerequisite	Description
Ad-hoc sub-process	Your process must contain an ad-hoc sub-process whose ID you can reference.

Create an Ad-Hoc Tools Schema connector task

1. Create a service task.
2. Apply the Ad-Hoc Tools Schema element template.
3. Configure the Ad-hoc sub-process ID to reference the element ID of the ad-hoc sub-process.
4. Configure a result variable or a result expression to map the connector results to process variables.

Tool Definitions

For more information on how to define tools, see the AI Agent connector documentation.

Response Structure

This connector returns a list of tool definitions in the following format. Individual tool definitions are modeled after the list tools response defined in the Model Context Protocol (MCP) specification, so you should be able to directly use the definitions with different LLMs or transform them into the required format for your AI integration.

{
  "toolDefinitions": [
    {
      "name": "GetDateAndTime",
      "description": "Returns the current date and time including the timezone.",
      "inputSchema": {
        "type": "object",
        "properties": {},
        "required": []
      }
    },
    {
      "name": "Download_A_File",
      "description": "Download a file from the provided URL.",
      "inputSchema": {
        "type": "object",
        "properties": {
          "url": {
            "type": "string",
            "description": "The URL to download the file from"
          }
        },
        "required": ["url"]
      }
    },
    {
      "name": "SuperfluxProduct",
      "description": "Calculates the superflux product (a very complicated calculation) given two input numbers.",
      "inputSchema": {
        "type": "object",
        "properties": {
          "a": {
            "type": "number",
            "description": "The first number to be superflux calculated."
          },
          "b": {
            "type": "number",
            "description": "The second number to be superflux calculated."
          }
        },
        "required": ["a", "b"]
      }
    }
  ]
}

You can either configure a result variable to contain the whole response or use a result expression to map parts of the response into your process.