Cannot read properties of undefined (reading 'match') on custom endpoints

[mitread]

Hello, I'm starting with LibreChat and it's really slick. Running into some problems:
I'm having issues hitting custom endpoints configured in LibreChat, I'm specifically getting the error in chat:

"Something went wrong. Here's the specific error message we encountered: An error occurred while processing the request: Cannot read properties of undefined (reading 'match')"

Running from source I find it's erroring out in

api/server/controllers/agents/client.js:741

const noSystemMessages = noSystemModelRegex.some((regex) =>
  agent.model_parameters.model.match(regex),
);

I found that it works just fine when hitting a ollama configured endpoint, but any custom endpoints throw an error. Logging the agent object shows two differently formatted objects:

ollama

{
  id: "ephemeral",
  instructions: undefined,
  provider: "ollama",
  model_parameters: {
    model: "llama3.2",
    baseUrl: "http://localhost:11434",
  },
  model: "llama3.2",
  tools: undefined,
  endpoint: "ollama",
  attachments: undefined,
  resendFiles: true,
  toolContextMap: undefined,
  maxContextTokens: 114750,
}

custom endpoint

{
  id: "ephemeral",
  instructions: undefined,
  provider: "openAI",
  model_parameters: {
    streaming: true,
    user: "685b0e253fd7aa11dec5cda3",
    apiKey: "ollama",
    configuration: {
      baseURL: "http://custom.example.end:7001/v1",
      defaultHeaders: {
      },
    },
  },
  model: "llama3-70b",
  tools: undefined,
  endpoint: "MLFlow",
  attachments: undefined,
  resendFiles: true,
  toolContextMap: undefined,
  maxContextTokens: 7200,
}

Obviously the model is missing from the model_parameters portion. I modified the code to skip the regex match, but then got other errors when making the request to the endpoint: 400 status code (no body). I can confirm I can hit the custom endpoint using a python script and the OpenAI package.

Here's my librechat.yaml file:

version: 1.1.4
cache: true
endpoints:
  custom:
    - name: "MLFlow"
      apiKey: "ollama"
      baseURL: "http://custom.example.end:7001/v1/"
      models:
        default: [
         "mixtral-8x22b","llama3-70b"
        ]
        fetch: false
      titleConvo: true
      titleModel: "current_model"
      summarize: false
      summaryModel: "current_model"
      forcePrompt: false
      modelDisplayLabel: "MLFlow"
      dropParams: ["stream","hide_sources","model"]
      directEndpoint: true
    - name: "Ollama"
      apiKey: "ollama"
      baseURL: "http://host.docker.internal:11434/v1/" 
      models:
        default: [
          "llama3.2","qwen3:4b"
          ]
        # fetching list of models is supported but the `name` field must start
        # with `ollama` (case-insensitive), as it does in this example.
        fetch: true
      titleConvo: true
      titleModel: "current_model"
      summarize: false
      summaryModel: "current_model"
      forcePrompt: false
      modelDisplayLabel: "Ollama"
mcpServers:
  test_mcp:
    type: streamable-http
    url: http://host.docker.internal:6278

Am I mis-configuring something?

Steps to reproduce:

1. Added an Open AI chat endpoint as listed above.
2. Reference that endpoint in the chat UI.
3. Type anything.

Relevant log output from docker:

LibreChat      | 2025-07-10 17:26:26 error: [api/server/controllers/agents/client.js #sendCompletion] Operation aborted Cannot read properties of undefined (reading 'match')
LibreChat      | 2025-07-10 17:26:26 error: [api/server/controllers/agents/client.js #sendCompletion] Unhandled error type Cannot read properties of undefined (reading 'match')
LibreChat      | 2025-07-10 17:26:26 error: Client title error: Run not initialized

Docker version:

ghcr.io/danny-avila/librechat-dev                      latest    34280f5146b2   3 days ago      1.1GB
ghcr.io/danny-avila/librechat-rag-api-dev-lite         latest    2991e83a1e7b   5 days ago      1.48GB

Source version: f4d97e16725a9e80cf4b7b47b4604a0f26901fee

Any pointers would be appreciated.

Thanks!