I've added a query tool with a 'knowledgebase' and a 'file' to lookup, added instructions to the assistant prompt on when to use it, but everytime it tries to use it, it returns the error "Couldn't Find Local Tool query". https://cdn.discordapp.com/attachments/1358374084102066316/1358374084693459064/image.png?ex=67f39bf3&is=67f24a73&hm=ab742fe1b6732da94c698695390a586bf487ca9fa926c30a65d9661455cbedc4&

Hi there! I see you're encountering the "Couldn't Find Local Tool query" error when your assistant tries to use your query tool. Let me help troubleshoot this issue! 👨‍💻 I notice that while I can see you've added a query tool with a knowledgebase and file lookup, I don't have specific details about your exact configuration. However, I can definitely help with common causes and solutions for this error. ## Possible Causes This error typically occurs due to several reasons: 1. **Tool Configuration Issues**: According to the information I found, this could be due to incorrect tool ID usage. One common mistake is using the tool name instead of the toolID when enabling the tool ([source](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1338953413886480425)). 2. **Response Format Problems**: The tool call response might not be in the correct format. The "result" field must be a single-line string without line breaks ([source](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1342188078437236916)). 3. **Variable Evaluation Issues**: The system might not be correctly evaluating variable paths before making the API call ([source](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1321375442644762625)). ## Solutions to Try Here are some steps to resolve this issue: ### 1. Check Tool Configuration Make sure you're using the toolID instead of the tool name when enabling the tool. The correct approach is:

// Use this (with toolID)
updateAssistant({ tools: [toolID] })

// Instead of this (with tool name)
updateAssistant({ tools: [toolName] })

### 2. Verify Prompt Instructions Explicitly integrate the tool in your prompt by clearly specifying when and how the assistant should use it. For example:

If the user wants to know about something, use the `query` function with the parameter 'query', which will contain the user's question to initiate the process.

Thanks for the help. Can you elaborate on the 1. Tool Configuration Issues I am using the vapi dashboard to edit the assistant

Hi there! Happy to elaborate on the Tool Configuration Issues, especially regarding toolID versus tool name when using the Vapi dashboard. 👨‍💻 Unfortunately, I don't have access to your specific dashboard configuration, but I can definitely help explain the common issues that might be causing your "Couldn't Find Local Tool query" error. ## Tool Configuration Issues in the Vapi Dashboard ### Using toolID vs. Tool Name One of the most common issues is using the tool name instead of the toolID when enabling a tool. According to information I found, when updating an assistant with a tool, using the tool name instead of the toolID can cause problems - specifically, it might create a new transient tool instead of enabling the existing one ([source](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1338953413886480425)). A toolID in Vapi is a unique string identifier that's used to identify a specific tool. It's often included in the initial request from Vapi and helps match the response with the corresponding tool call. ### Incorrect Tool Configuration Structure When configuring a function in the dashboard, the

toolId

property should not be present in the

tools

array. If it is, you might see error messages like:

"message": ["assistant.model.each value in tools.property toolId should not exist"]

The correct tool configuration for a function should include: -

type

(e.g., "function") -

functionName

(the name of your function) -

async

(boolean value) - Any required parameters ### Dashboard Limitations and Bugs There are some known issues with the Vapi dashboard interface: - Adding headers to tools directly on the dashboard may not work due to a known bug - Some users have reported that the dashboard UI doesn't always save settings correctly - The dashboard only displays the 100 newest assistants, so if you have more, the one you're looking for might not be visible ### Tool Type Specification

Thank you for reaching out to us. We've received your message and will get back to you as soon as possible.

@Vapi Support Bot Hey, I have the same issue, and changing the name of the tool to it's id didn't help

Hi there! I understand you're experiencing the same "Couldn't Find Local Tool query" error and that changing the tool name to its ID didn't resolve the issue. Let me suggest some additional troubleshooting steps! 🔍 Unfortunately, I can't see your specific configuration, but I can definitely suggest other potential solutions beyond just changing the tool name to its ID. Here are some additional approaches you could try: ### 1. Check Response Format Ensure your tool call response is in the correct format:

json
{
  "results": [
    {
      "toolCallId": "<your-tool-call-id>",
      "name": "<your-function-name>",
      "result": "Success" // Must be a single-line string without line breaks
    }
  ]
}

According to [documentation](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1342188078437236916), a common mistake is using line breaks in the "result" field or using the tool ID instead of the tool call ID. ### 2. Verify Tool Configuration Structure When configuring a function in the dashboard, make sure: - The

type

field is correctly set (e.g., "function") - The

functionName

is properly defined - The

async

parameter is set appropriately - The

toolId

property is NOT present in the

tools

array ### 3. Check Variable Evaluation According to [troubleshooting information](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1321375442644762625), the system might not be correctly evaluating variable paths before making the API call. Ensure that variable substitution is set up correctly. ### 4. Try Different Model or Cluster Some users have reported success by switching to a different model like "gpt-4o" or using a different cluster instead of the real-time cluster, as mentioned in [this source](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1337465172263960637). ### 5. Properly Create the Tool Call

@Vapi Support Bot Could the issue be that I'm using a transferCall function? I've verified its set up properly, however the "Couldn't Find Local Tool transferCall" is still displayed

I see you're now experiencing a "Couldn't Find Local Tool transferCall" error rather than the query tool error we were discussing earlier. This is a different issue related to the

transferCall

function. Could you please share more details about how you've configured the

transferCall

function? Specifically: 1. Are you using the Vapi dashboard or API to configure it? 2. Have you checked if there's a conflict with any server events, particularly the 'phone-call-control' event? 3. What format are you using for the phone number in the

forwardingPhoneNumber

parameter? 4. Are you using SIP for call transfers, and if so, is your system configured to handle SIP REFER messages? This information would help me provide more targeted troubleshooting steps for the

transferCall

function specifically. Sources: - [What is the transferCall function in Vapi?](https://docs.vapi.ai/call-forwarding) - [What are potential issues with the transferCall function in Vapi?](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1330092161835733013) - [What are the solutions for the "Couldn't Find Local Tool transferCall" error in Vapi?](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1339871158975336498)

@Vapi Support Bot I've used api to initially create it, and it worked flawlessly before, but after this Friday (without any changes) it stopped working. I've tried changing the agent settings in the dashboard. The only server messages I currently have is transfer-destination-request. I've used E164 format for all phone numbers in my destinations, and double-checked they are correct.

Thank you for providing those details about your

transferCall

function setup. Since it was working before Friday and suddenly stopped without any changes on your end, this sounds like it could be related to a recent platform update or a backend issue. A few more questions to help narrow down the problem: 1. Have you checked the call logs for any error messages beyond the "Couldn't Find Local Tool transferCall" that appears in the chat? 2. When you say you've tried changing agent settings in the dashboard, which specific settings did you modify? Did you try removing and re-adding the

transferCall

tool? 3. Are you using a custom LLM or the default Vapi-provided models? If custom, have you ensured the

transferCall

tool is included in your custom LLM tools? 4. Have you tried testing with a simple call transfer to a single destination to rule out any issues with multiple destinations or complex routing logic? Sources: - [What are potential issues with the transferCall function in Vapi?](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1339614362742165606) - [What are the solutions for the "Couldn't Find Local Tool transferCall" error in Vapi?](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1339871158975336498) - [How to configure the transferCall function in Vapi?](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1335181348285054996)

@Vapi Support Bot I tried removing and re-adding the tool to the agent, I'm using basic vapi-provided llm (Gpt-4o-cluster), I also tried using a cluster-mini which didn't solve the issue.

Hi there! I see you're experiencing the "Couldn't Find Local Tool transferCall" error despite having previously working configuration. Let me help troubleshoot this issue! 🔍 Unfortunately, I can't determine exactly what might have changed on the platform since Friday, but I can suggest several troubleshooting steps specifically for the transferCall function. ## Potential Solutions to Try: ### 1. Remove the "phone-call-control" server event If you have the "phone-call-control" server event enabled, try removing it from your assistant. According to [information I found](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1330092161835733013), when this event is enabled, you handle call transfers instead of Vapi, which can cause conflicts. ### 2. Ensure the transferCall tool is included in your model configuration Make sure the transferCall tool is properly defined in your model's

tools

array. For GPT-4o, you need to ensure the tool is correctly configured with the appropriate parameters. ### 3. Check the response format The correct format for tool call responses should be:

json
{
  "results": [
    {
      "toolCallId": "<your-tool-call-id>",
      "name": "<your-function-name>",
      "result": "Success" 
    }
  ]
}

The

result

must be a single-line string without line breaks, and you should use the tool call ID instead of the tool ID, as mentioned in [this source](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1342188078437236916). ### 4. Try a different approach to tool configuration Since you mentioned you've already tried removing and re-adding the tool, consider: - Using the API directly instead of the dashboard for configuration - Not using "transferCall" as the tool call name (this is a [recommended tip](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1305244632124227654)) - Setting

assistantOverrides

correctly if you're using silent transfers

No it didn't work. I tried creating a new query tool with the same parameters but still encounter the same issue

Thank you for your patience, will get back to you when available

Hey KosKrit, could you share either the call ID or assistant ID so I could take a look?

Thank you. It works now. Even the older assistants work, so I suppose it was fixed from your end. Do you still need the assistant ID for troubleshooting?

Is working for you? Then, Assistant ID is not required.