In Chat Logs and Api, it shows that the tool has been successfully called. But it has been really not called.

Check your custom tool's server endpoint and response format. Make sure your tool's server is accessible and returns a response like this:

json
{
  "results": [
    {
      "toolCallId": "YOUR_TOOL_CALL_ID",
      "result": "Your result here"
    }
  ]
}

If the endpoint is unreachable or the response format is incorrect, Vapi may log the tool call as "successful" even though it wasn't actually executed. Double-check your server URL and test it directly to confirm it works as expected. [See the official docs for details.](https://docs.vapi.ai/tools/custom-tools#server-response-format-providing-results-and-context) Sources: - [Custom Tools – Server Response Format](https://docs.vapi.ai/tools/custom-tools#server-response-format-providing-results-and-context) - [Custom LLM Tool Calling Integration](https://docs.vapi.ai/customization/tool-calling-integration#c-custom-tools)

Check that tool responses have correct toolCallId, variables are properly passed, system prompts instruct tool usage clearly, and review webhook and call logs for errors..

so i will need to make other flow than voice one for chat?

You should create a separate assistant flow with proper toolCallId handling, clear system prompts, and valid JSON responses, as voice and chat flows require different configurations..

that didnt help, as far as i can see custom tools configs are singular according to your docs. other than custom tools what configs are different? can you share some links for it?

Hey, can you share your assistant ID so I could take a look?

Having this same issue

@Shubham Bajaj 7d40c99a-7a9c-4ba8-ae02-ee117a657da9 assistant id

Hey @hxrdxk @kitten can you guys try now and share the session_id if this issue appears for you again?