Configuration Tab of a Tool

When creating or editing a Tool, two tabs are shown. The first tab, titled Configuration, is where you define how a Tool connects to external services and what inputs it requires.

This tab is divided into three collapsible panels:

• Scope
• Information
• Parameters

Scope

This panel establishes the connection between the Tool and an external service. Globant Enterprise AI Lab supports dynamic data retrieval via OpenAPI, allowing Tools to interact with external APIs.

You must choose one of the two available input methods:

• URL: Use this option when you want to connect to an API through a direct endpoint. In this case, you need to provide the complete API URL and any necessary authentication. Then, paste the full OpenAPI URL into the Open API field.

• JSON: Use this option when you have the OpenAPI specification available in JSON format. Paste the full content in the Open API Json field, allowing Globant Enterprise AI Lab to automatically identify available endpoints and operations.

Information

This section defines how the Tool is displayed and identified.

It includes the following fields:

• Tool Name (required): Allows Globant Enterprise AI Lab to correctly identify and route the operation defined in the OpenAPI specification.

  • If the operationId is unique, use the format: operationId + method (e.g., AI_studio_agent_get).
    Note: Although an OpenAPI specification can contain multiple operations or endpoints, each tool created in Globant Enterprise AI Lab represents only one of them. If you need to use more than one endpoint from the same API, you will need to create a separate tool for each.

  • If the same operationId appears in multiple paths, use the format: operationId + normalized path + method (e.g., AI_studio_agent_agents__agentid_get).

• Description (optional): Provide a brief description of what the Tool does. This helps to understand its functionality.

Parameters

The Parameters section defines the inputs your Tool requires to function correctly. These inputs are used to process data and execute external API requests.

If no parameters have been added yet, the interface will display the message:No parameters configured yet.

To add a parameter, click on + Add new parameter. Next, fill in the following fields:

• Key: Identifier used to reference the parameter. It must match the exact name of the parameter as defined in the OpenAPI specification. For example, if the endpoint includes parameters such as "queryText" and "provider", those same names must be used as keys in the tool configuration to ensure the values are correctly mapped at runtime.
• Description: Brief explanation of the parameter’s role.

• Type: Defines the kind of parameter you are configuring. You can choose from the following:

  • Application: Used to pass dynamic values at runtime, typically based on user input or the specific execution context. For example, a Tool that sends an email may receive the recipient address and message content as application parameters. These values are resolved by the LLM during execution and are often sent as query parameters, particularly in GET requests.
    This type is especially useful when the OpenAPI description is not detailed enough for the LLM to infer what values are expected, or when additional guidance is needed to ensure the tool behaves correctly during execution.
  • Config: Used to define fixed values such as API tokens or credentials to be configured per project. They are usually sent in headers and remain static during execution. This is the only type of parameter that you can view and modify directly. Optionally, a Config parameter can extract its value from a Secret to enhance security.
  • Context: This type of parameter allows the Tool to automatically retrieve values from the runtime session, such as user, project, or organization information. These parameters do not require manual configuration, since their values are dynamically resolved during execution.
    Some commonly used context keys include:
    • API_TOKEN: Retrieves the token associated with the current project. If no token is found, the system uses a default value.
    • USER_ID, USER_NAME, USER_EMAIL: Provides information about the user who initiated the request.
    • ORGANIZATION_ID, PROJECT_ID, ORGANIZATION_NAME, PROJECT_NAME: Identifies the organization and project context.

  If the key matches the context variable name expected by the tool, the Value field can be left blank. The list of available context keys is predefined in AI Lab and can be queried through the API when needed.

  In addition, the Secret setting has no effect on the context parameters and can be left unselected.

• Value: Sets the default or fixed value for the parameter.

  • For Config parameters, this field is required and determines the static value the tool will use. You can enter a fixed literal value (e.g., a base URL or timeout) or reference a Secret by specifying its name. For example, if you enter "My Secret API Key" in this field, the Tool will look for a Secret with that exact name to retrieve the actual value securely.
  • For Context parameters, this field is optional and only needed if the key and context variable differ.
  • For Application parameters, this field is typically left blank, since values are provided dynamically at runtime.

• Data Type: Select the data type from:

  • Boolean
  • Decimal
  • JSON
  • Number
  • String

• Required: Enable this option if the parameter is mandatory.

• Secret: Enable this option if the parameter value contains sensitive data. This setting is only applicable for parameters of Type Config. See How to create a Secret.

To remove a parameter, click on the trash can icon next to it. Then confirm the action by clicking on the check icon that appears.

Availability

Since April 2025 release.