Call Agent
A Call Agent Tool allows one Agent to invoke another Agent as a Tool within a Scenario. This enables you to build modular, composable Agent architectures where a parent Agent delegates specific tasks to specialised child Agents — keeping each Agent focused on a well-defined responsibility.
When to Use
Use a Call Agent Tool when:
A Scenario requires capabilities handled by an existing Agent
You want to reuse a specialised Agent across multiple parent Agents
A complex Workflow is better served by chaining Agents together rather than building a single monolithic Agent
You want to keep Agents modular and independently maintainable
Creating a Call Agent Tool
From the My Tools page, click Create and select Call Agent Tool from the Create tool from options at the top of the page.
The creation flow has four sections:
About Tool
This section defines the Tool's identity as it will appear to the parent Agent during reasoning.
Tool name
Yes
A descriptive name that clearly conveys which Agent this Tool invokes and what it does — for example, callOrderAgent or invokeFinancialAssistant
Description
Yes
A plain language explanation of what the called Agent does and when the parent Agent should invoke it. The parent Agent uses this description to decide whether to call this Tool for a given user request
Notes
No
Internal notes for documentation purposes — not used by the Agent during reasoning
💡 Note: A specific and well-defined description improves AI responses in the parent Agent. Be explicit about what kinds of questions or tasks the called Agent is designed to handle.
Input Parameters
Input parameters define the information passed from the parent Agent to the called Agent at invocation time.
Each parameter has the following fields:
Parameter name
A descriptive identifier for the input — for example, userQuery or customerId
Type
The data type — such as String, Integer, or Boolean
Description
Explains what the parameter represents and how the parent Agent should populate it
Click Add parameter to define additional inputs as needed. Reference parameters in templates using the format ${<parameter-name>} — for example, ${userQuery}.
💡 Note: The parameters defined here should align with what the called Agent expects to receive in order to process the request correctly.
Output
Defines the structure of the response returned by the called Agent back to the parent Agent.
Output type name
The type of output returned — typically Object for a single response
Description
A plain language description of what the called Agent returns — for example, "the Agent's response to the user query"
Object structure
A JSON definition describing the structure of the returned response
Example output structure:
Mark output as artifact — Enable this when the called Agent returns a file or artifact such as a document, image, or audio file rather than structured data.
💡 Note: If you are unsure of the exact output structure, run the Tool using the Test panel first and apply the returned structure automatically.
Configuration
The Configuration section is where you select the specific Agent this Tool will invoke. This links the Tool to the target Agent in your Dataworkz environment.
💡 Note: The Agent you want to call must already exist in your environment before creating the Call Agent Tool. If it does not exist yet, create and configure the child Agent first, then return to create this Tool.
Testing the Tool
Use the Test panel on the right side of the page to verify the Call Agent Tool is invoking the target Agent correctly and returning the expected output before saving.
⚠️ Warning: Always test the Call Agent Tool before adding it to a parent Agent Scenario. If the called Agent is not configured correctly or the parameters do not align, the parent Agent will fail when attempting to invoke it.
Saving the Tool
Once all sections are complete and the Tool has been tested, click Save & Close to save and return to the My Tools list, or Save to save without closing.
💡 Note: Once saved, the Call Agent Tool is available to be referenced in any Scenario of any parent Agent within your Dataworkz environment, making the called Agent reusable across multiple Workflows.
Last updated