userIdQuery parameter value is always set to "true"

[jcbartle]

What happened?

When using a custom endpoint, user is always included in the request body, even if userIdQuery is set to false.

Here is my config:

  custom:
  - name: "Azure AI Foundry"
    apiKey: "user_provided"
    baseURL: "https://xxxxx.xxxxxxxx.xxx/xx/xxxxxxxxxxxxx"
    directEndpoint: false
    headers:
      User-Identifier: "{{LIBRECHAT_USER_EMAIL}}"
    models:
      default:
        - "model_id"
      fetch: false
      userIdQuery: false
    titleConvo: true
    titleMethod: "completion"
    titleModel: "gpt-4.1-nano"
    summarize: false
    modelDisplayLabel: "model_display_name"

Checking the actual API call request body as revealed in Azure Monitor shows that the user parameter is still included in the body. This causes the following error:

Extra parameters ['user'] are not allowed when extra-parameters is not set or set to be 'error'. Set extra-parameters to 'pass-through' to pass to the model.

Obviously, setting the extra-parameters header fixes this. However, this shouldn't be required if the userIdQuery parameter value were being honored from the YAML file.

Version Information

LibreChat v0.7.9-rc1

Steps to Reproduce

1. Set userIdQuery to false and run a query.
2. Set userIdQuery to true and run a query.
3. Check the API call request bodies.
4. They will be identical.

What browsers are you seeing the problem on?

Microsoft Edge

Relevant log output

Looking at the log content, there didn't appear to be anything related to or referencing this setting.

Screenshots

No response

Code of Conduct

• I agree to follow this project's Code of Conduct

[danny-avila]

userIdQuery is not a recognized parameter. You should use dropParams

[jcbartle]

I can confirm that this method works just fine.

[jcbartle]

Hi, @danny-avila. Thanks for the quick response, and I think that's a fine solution.

Can the documentation be updated to reflect this?

https://www.librechat.ai/docs/configuration/librechat_yaml/object_structure/custom_endpoint#models

[danny-avila]

userIdQuery is for the v1/models request to populate the list of models dynamically. I believe the documentation is accurate as is.