Tool Calling Problem with Supervisor-Worker Type of Flow

[eastnbull]

Bug Description

Hi everyone!

I’m encountering an issue with the ToolCalling Agent, and I’d appreciate your help.

I’ve created two flows: one with a single agent, and another with two agents working together. The single-agent flow successfully calls an LCToolComponent, triggering a search and I can see the output from the endpoint.

However, the multi-agent flow isn’t working as expected. The idea behind this setup is to use a coordinator/supervisor email agent that calls a ChatGPT agent, which then triggers the LCToolComponent for the tool call. While the ChatGPT agent is successfully triggered, the tool call doesn’t happen in this multi-agent setup.

I couldn’t find any documentation or material that covers this specific issue, so I’d be grateful for any insights or advice.

Thank you in advance for your support!

Reproduction

Pls take a look to my attached flow files. You will see your self.

Expected behavior

Worker agent can call a tool when it is initiated by a Supervisor

Who can help?

No response

Operating System

Macos Sequoia 15.2

Langflow Version

1.1.1

Python Version

None

Screenshot

No response

Flow File

Multi Agent with ToolCalling (1) (1).json
Single Agent with Tool Calling (1) (2).json

[dosubot]

In a multi-agent setup in Langflow, the ToolCallingAgent should be able to trigger tools like the LCToolComponent when properly configured. Here are a few things to check and consider:

1. Tool Mode Activation: Ensure that the Tool Mode is enabled for the components involved in the tool calling process. This is crucial for allowing agents to use tools within a multi-agent flow ^[1].

2. Correct Connections: Verify that the connections between the supervisor agent, the ChatGPT agent, and the LCToolComponent are correctly established. The Tools port of the agent should be connected to the Toolset port of the LCToolComponent.

3. Compatible Language Model: Make sure that the language model used supports tool calling. If the model does not support this feature, a NotImplementedError might be raised, preventing the tool call from occurring ^[2].

4. Error Handling: Check for any error messages or logs that might indicate why the tool call is failing. The handle_on_tool_error function can provide insights into any errors that occur during the tool calling process ^[3].

5. Flow Configuration: Review the flow configuration to ensure that all necessary components are correctly set up and that there are no missing configurations that could affect the tool calling process.

By addressing these points, you should be able to identify and resolve the issue with the tool call in your multi-agent flow.

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Other}