Adding tool and toolkit for an AI Agent

A tool is a program that helps an AI agent perform specific tasks or actions, while a toolkit is a collection of such tools.
You can create a custom tool from the available types and add it to the AI agent. When adding a tool or toolkit, make sure that the agent supports it.

To add a tool or toolkit

Log in to HelixGPT Agent Studio.
Click Tools.
The Tools screen is displayed as shown in the following screenshot:

Click Add tool/toolkit.
The Add Tool dialog box is displayed.

In the Add Tool dialog box, enter the following values:

Field name	Description	Screenshot
Name	Ener the tool name.
Type	Select the tool type from the available options. All tool types are available when you create a new tool or toolkit. However, only applicable tool types appear when you edit an existing tool or toolkit. For more information, see Tool types and usage.
Description	Enter a valid description for the tool.
Configuration	Define the tool configuration in the JSON format.
Connection	If applicable, select a connection from the available options.

Click Save.

To edit an existing tool or toolkit

Click on a tool or toolkit from the Tools list.
The Edit Tool dialog box as shown in the following screenshot is displayed:
Update the values.
Click Save.

Tool types and usage

Tool types define the specific functionalities that AI agents can perform. They help agents process tasks such as retrieving data, generating content, or integrating with other systems. Different tool types are used to enhance the agent's capabilities and improve task efficiency.

For example, a Knowledge tool helps an AI agent fetch relevant information from a knowledge base.

The following table describes the available tool type and its usage:

Tool type	Usage
BAR ITSM	Interact with IT service management systems, performing tasks such as creating incidents, updating tickets, and retrieving service requests.
BWF Case Tool	Create, update, and manage cases in BMC Helix Business Workflows, streamlining case management by automating routine tasks.
DWPC Tool	Interact with BMC Helix Digital Workplace Catalog, managing service requests, approvals, and catalog items efficiently.
DWP Event Tool	Capture and process events in BMC Helix Digital Workplace, triggering appropriate actions based on user activity or system events.
DWP Profile Search Tool	Search and retrieve user profile information from BMC Helix Digital Workplace, personalizing responses and streamlining user interactions.
DWP Service Health Tool	Check and display the health status of services in BMC Helix Digital Workplace, helping users stay informed about service availability and performance.
Foundation	Access and manage foundational data, such as user profiles, organizations, and locations, ensuring accurate information retrieval and process execution.
IS Process Tool	Perform processes in BMC Helix Innovation Studio, efficiently automating workflows and managing business processes.
ITSM	Interact with IT service management systems, performing tasks such as creating incidents, updating tickets, and retrieving service requests.
Knowledge Tool	Search, retrieve, and suggest relevant knowledge articles, improving issue resolution by providing accurate information quickly.
MFS Tool	Interact with the Multi-Form Search (MFS) capability, retrieving relevant data by searching across multiple forms and applications.
Microsoft Teams	Facilitate communication and collaboration within Microsoft Teams, enabling users to create incidents, check ticket status, and interact with support teams.
Salesforce Knowledge Tool	Search and retrieve knowledge articles from Salesforce, providing relevant information to efficiently resolve user queries.
ServiceNow Knowledge Tool	Search and retrieve knowledge articles from ServiceNow, providing accurate information to quickly resolve user queries.
Supervisor Agent Handoff Tool	Escalate complex issues to a human supervisor, ensuring a seamless handoff for better issue resolution and user satisfaction.
xMatters	Trigger notifications and alerts through xMatters, ensuring a timely communication and incident resolution.
Live Agent Tool	Enables users to connect with a human agent directly from the virtual assistant when additional support is needed, ensuring a seamless and efficient handoff for quicker issue resolution.

Configuring the IS Process Tool

The IS Process Tool triggers a process in BMC Helix Innovation Studio. The tool initiates and manages IS workflows, utilizing a structured configuration that outlines how processes are executed and how results are retrieved.

Configuring the IS Process Tool involves specifying the process name, input parameters, and expected output variables, enabling AI agents to interact dynamically in the workflow. For a detailed use case, see Integrating GitHub with BMC HelixGPT

To configure the IS Process Tool

Use the following configurations when configuring the IS process tool.

Log in to HelixGPT Agent Studio.
Click Tools.

Click Add tool/toolkit.
In the Add Tool dialog box, enter the following values:

Field name

Description

Name

Enter the tool name.

Type

Select the IS process tool from the drop-down list.

Description

Enter a valid description for the tool.

Configuration

Define the tool configuration in the JSON format. Specify the following settings

Setting		Description
process_name		The name of the IS process to be invoked.
input_parameters	(Optional) The input parameters required to invoke an IS process. The list can be empty if your process doesn't require any inputs.
	name	The name of the input parameter. This must match the name of the variable in the process.
	description	The description of the parameter. This helps the LLM understand how to set the input value. For example: If the parameter is Name, write the description as: "Full name of the user. This is a required parameter."
	type	The string value of a Python type. This informs the LLM about the expected input format. For example: Use "int" for integers, "str" for strings
	optional	(Optional) The boolean flag that indicates whether the parameter is optional. Default value: False
output		(Optional) The list of process variables that must be returned to LLM as a process result. If the output value is empty, all the process variables are presented as output.
request_overlay_group		(Optional) Request-Overlay-Group header. It's recommended not to set this value if it is not required in the IS process. Default value: 1 (custom mode) Use 0 for Base mode. For more information about the customization layers, see Customization layer.

Sample configuration:

Click to view the sample configuration

{
  "process_name":"com.example:GetWeatherForecast",
  "input_parameters":[
      {
        "name":"city",
        "description":"The city for which to get the weather forecast, this is a required parameter",
        "type":"str"
      },
      {
        "name":"units",
        "description":"Unit of measurement for the forecast, valid values are `metric` or `imperial`, this is an optional parameter that the user can specify, you should default to 'metric'",
        "type":"str",
        "optional": true
      }
   ],
  "output":["result"],
  "request_overlay_group": 1
}

Providing a reference link to the user

The IS process tool can optionally provide a user with a reference link to the data it returns. This reference link serves as a source for the returned data, enabling end users to verify the information, explore further details, and take follow-up actions based on it.

To provide the reference links to the user, specify the following output variables in the Configuration field of the IS process tool.

Output variables	Description
web_url	The URL of the page that references the data
title	The title of the page that references the data

Supported Process Variables

The following table lists the supported variables for input and output parameters to be mentioned in the IS process tool configurations.

Variable type	Supported	Python type	Accepteble values / notes
Attachment	❌ No
Boolean	✅ Yes	bool str int float	(case insensitive) true false "true" "false" 1 0 1.0 0.0
Date	✅ Yes	str int	String format must match the platform definition
Date/Time	✅ Yes	str int	String format must match the platform definition
Decimal	✅ Yes	str int float
Floating	✅ Yes	str int float
Integer	✅ Yes	str int
Object	✅ Yes		Any value
Record	✅ Yes	dict	Example:
Selection	✅ Partial	str int	Support only for the selection integer value, not the selection option name
Text	✅ Yes	str	Any value will be converted to a string by IS
Time	✅ Yes	str int	String format must match the platform definition
List	✅ Yes	list	The list of values and the value characteristics depend on the value type (see above)