Disable File Uploads Per User

[jmariquit-tvlk]

We already have the capability to disable file uploads on a per model basis. (Thank you very much for that one!)

Are there plans on of having the ability to disable it per user as well (either via an npm CLI or as a config parameter):

• perhaps as a role/attribute in the OpenID claim
• or a property as part of the user profile (so that we can have two types of users: chat users versus power users who attaches files for enhanced context) -- like a 'allowUpload' property that can be set via npm CLI

My apologies in advanced if this has been brought up before -- my search-fu skills is crap

[nth347]

hi @jmariquit-tvlk , could you please tell how you configured file upload controls (file types, file size)?

[jmariquit-tvlk]

here you go: https://www.librechat.ai/docs/configuration/librechat_yaml/object_structure/file_config

let me know if you can't get the yaml figured out

also, do remember that file uploads will be charged against your tokens (i.e. OpenAI tokens, etc.) -- including Librechat's own token credit balances

[nth347]

hello @jmariquit-tvlk,

here is my config

fileConfig:
  endpoints:
    assistants:
      fileLimit: 5
      fileSizeLimit: 10  # Maximum size for an individual file in MB
      totalSizeLimit: 50  # Maximum total size for all files in a single request in MB
      supportedMimeTypes:
        - "image/.*"
        - "application/pdf"
    openAI:
      disabled: true  # Disables file uploading to the OpenAI endpoint
    google:
      disabled: true
      fileLimit: 1
      fileSizeLimit: 1
      totalSizeLimit: 3
      supportedMimeTypes:
#        - "image/.*"
        - "application/pdf"
    default:
      totalSizeLimit: 20
    YourCustomEndpointName:
      fileLimit: 2
      fileSizeLimit: 5
  serverFileSizeLimit: 100  # Global server file size limit in MB
  avatarSizeLimit: 2  # Limit for user avatar image size in MB

For google, I set disabled to true and updated supportedMimeTypes to application/pdf. However, I was still able to upload a picture in jpeg format and ask about it in the chat using Google Gemini. Did I miss something?

[jmariquit-tvlk]

according to the docs, setting the disabled: true under that endpoint (in this case Google), should override any other file-related settings. You might have to raise this as a bug.

in the meantime, try commenting out the other keys so that you have something like

fileConfig:
  endpoints:
    google:
      disabled: true
    openAI:
      disabled: true

[nth347]

hi @jmariquit-tvlk , I used the minimized config as you suggested, but it didn't work anyway.

fileConfig:
  endpoints:
    google:
      disabled: true
    openAI:
      disabled: true

Hi @danny-avila, could you please share your opinion before I raise an issue?