Configure Tool Nodes in a Workflow

Tool nodes provide direct access to common tools that workflow agents can use to perform specific tasks. You can use Tool nodes to generate deep links, process chat attachments, identify intent changes, and retrieve information about the signed-in user. The fields available for configuration depend on the selected tool type.

Configure a Tool Node

1. Go to AI Agent Studio.
2. From the Workflows tab, edit a workflow and add a Tool node to it.
3. Enter a name. The code is automatically populated based on the name.
4. Enter a description for the Tool node.
5. Select an error handler node to define failure handling.
6. In the Type field, select one of the following tool types:
   • Deep Link
   • Connector
   • Chat Attachments Reader
   • Intent Change Indicator
   • User Session
7. Configure the fields for the selected tool type. The configuration fields vary depending on the tool type that you select.

   Tool Type	Description
   Deep Link	Enables workflows to reference and execute an existing Deep Link tool without recreating it. This tool is used to generate context-aware navigation links to Oracle Fusion Cloud Applications; for example, links to business objects or transactions. This tool mirrors the functionality of Deep Link usage in supervisor agents. It references preexisting Deep Link tools, returns a URL that points to a specific business function, and enables seamless navigation from the workflow to the application UI. Select Deep Link as the type and enter these details: Family: Select the family of products to associate this tool node with. Product: Select the product you're working with. Tool: Select the tool to use in this step. Recommend: Select Recommend, add a query and then select Go. This field searches for existing Deep Link tools that match the specified query and returns relevant recommendations.
   Connector	Enable adding an existing connector as a tool to connect to internal or external data sources. Tool: Select the connector tool to use. Function: Select the function to use from the connector.
   Chat Attachments Reader	Enables workflows to process files uploaded during a chat session. The output is the extracted content, provided as plain text in Markdown or Base64 format. This tool is also known as Multifile Processor. Select Chat Attachments Reader as the type and enter these details: Conversation Identifier: Specify the variables to uniquely identify conversations and apply them to the attachments. Type {{ to add a variable, or use expressions to specify the conversation identifier. For more information, see Expressions. Parse Files: Select to enable extraction of readable content or metadata. This option is on by default and is best for most text-based use cases. Make Raw File Content Available (in base64 format): Select to get the original bytes of the attachment for downstream processing or storage. For example: Pass files to downstream APIs that accept Base64, like vLLM. Preserve the original file for storage when the integration expects Base64. Process nontext or hard-to-parse files like images, scans, and proprietary formats. Re-create the file later, for example, by decoding Base64 back into the original binary in another step.
   Intent Change Indicator	Enables workflows to reset or segment the chat history based on the intent. Use expressions to specify the identifier used to detect intent changes. For more information, see Expressions.
   User Session	Enables adding the User Session tool as a node and returns the data from the tool relevant for the signed in user, such as PersonNumber or Username. Select User Session as the type.

8. Publish the workflow.

When the workflow runs, the Tool node runs the configured tool and returns the output generated by the selected tool. The output can be used by downstream workflow nodes to continue processing.