Hi, I am connecting to my assistant through WebSocket. When I request for the WebSocket URL from VAPI servers, I send under assistantOverrides the clientMessages, with "tool-calls-results" and other control messages I need to receive. For some reason, I receive all the control messages I need - but not "tool-calls-results". Under "Call Logs" in the website it can be clearly seen that the assistant receives the response from my API accurately, and he also provides speech-answer for that. I already tried it tens of times to figure out if the problem is in my code. One call id example: 019ad950-5927-7cc8-9769-6fdf6a0533e1 Note: I also tried to wait for longer time before I hang up the call. Still - never received this specific control message. It's important for me to receive it, as I need to show the results as a component in the UI. Thanks!

Suggestion: Double-check that your

clientMessages

array in

assistantOverrides

includes the exact message type expected by Vapi, which may be

"tool.completed"

instead of

"tool-calls-result"

due to recent updates.

json
"clientMessages": ["tool.completed", ...]

Review the [Server events documentation](https://docs.vapi.ai/server-url/events) to confirm the correct message type and update your configuration accordingly. Sources: - [Server events documentation](https://docs.vapi.ai/server-url/events) - [Changelog April 26, 2025](https://docs.vapi.ai/changelog/2025/4/26)

Message marked as helpful by @tomer_r97! 🎉

Hi tomer_r97, To address various challenges with tool calls and results, you can leverage the latest updates and functionalities. Here are a few key points: 1. **Metadata in ToolCallResult**: You can add optional metadata to tool call results and messages for additional context. More details can be found [here](https://docs.vapi.ai/changelog/2025/4/26). 2. **Client-Side Tool Calls**: If you're handling tool calls on the client-side, ensure that you're listening for

tool-calls

events with the Web SDK and updating the UI as necessary. Remember that client-side tools do not send results back to the model. For setting up client-side tool calls, refer to the [Client-side tools documentation](https://docs.vapi.ai/tools/client-side-websdk). 3. **Handling Errors**: The [Server Message Specification](https://docs.vapi.ai/api-reference/webhooks/server-message) provides detailed descriptions of various error scenarios that might occur during tool calls, which can help diagnose issues with your setup. For any troubleshooting process, make sure your implementation aligns with these guidelines and support resources. This should help in effectively utilizing tool calls within your application.

tool.completed is not in the referenced documentation AFAIK

Hello