[Custom Endpoints] fileConfig.endpoints.custom array structure not supported in release images (incorrect config)

[youvnor]

What happened?

Describe the bug

When attempting to configure a custom endpoint (e.g., Perplexity) via fileConfig.endpoints.custom in librechat.yaml, the application throws a Zod schema validation error if custom is defined as an array (the documented, correct method).

Zod validation error:

{
  "code": "invalid_type",
  "expected": "object",
  "received": "array",
  "path": ["fileConfig", "endpoints", "custom"],
  "message": "Expected object, received array"
}

Version info
LibreChat Version: latest (ghcr.io/danny-avila/librechat-dev-api:latest)
Deployment: Docker Compose (official docs for Linux)
OS: Debian bookworm

Relevant config (redacted):

fileConfig:
  endpoints:
    custom:
      - name: "Perplexity"
        apiKey: "${PERPLEXITY_API_KEY}"
        baseURL: "https://api.perplexity.ai/"
        models:
          default: ["sonar"]
          fetch: false
        modelDisplayLabel: "Perplexity"

ENDPOINTS=openAI,google,anthropic,custom,agents
PERPLEXITY_API_KEY=xxxx

I opened a discussion about it:
#7781

Additional context

At the moment I am unable to build the project from source in my current environment, but I will provide feedback as soon as I am able to test in a fresh environment. Should you require tests or logs from a newly built setup, please let me know and I'll prioritize that.

Version Information

ghcr.io/danny-avila/librechat-dev-api latest ec732384bd62 3 days ago 1.12GB
ghcr.io/danny-avila/librechat latest 17cfdb1df083 4 weeks ago 1.02GB
ghcr.io/danny-avila/librechat-rag-api-dev-lite latest e7da8e6aca0f 4 weeks ago 1.31GB

Steps to Reproduce

1. Use ghcr.io/danny-avila/librechat-dev-api:latest image.
2. Configure fileConfig.endpoints.custom as array (per documentation).
3. Start the server.
4. Observe Zod validation error.

What browsers are you seeing the problem on?

No response

Relevant log output

{"level":"error","message":"Invalid custom config file at /app/librechat.yaml:\n{\n  "issues": [\n    {\n      "code": "invalid_type",\n      "expected": "object",\n      "received": "array",\n      "path": [\n        "fileConfig",\n        "endpoints",\n        "custom"\n      ],\n      "message": "Expected object, received array"\n    }\n  ],\n  "name": "ZodError"\n}"}

Screenshots

No response

Code of Conduct

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

[danny-avila]

Hi @youvnor

You are confusing the endpoints config with the fileConfig object structure.

They are slightly different in schema, but serve very different purposes. You do not define custom endpoints under fileConfig.

Here's a working YAML file using both, note the distinctions. fileConfig is for setting file parameters (and note that it only supports defining blanket settings to custom instead of individual custom endpoints, i.e. Perplexity

endpoints:
  custom:
    - name: "Perplexity"
      apiKey: "${PERPLEXITY_API_KEY}"
      baseURL: "https://api.perplexity.ai/"
      models:
        default:
          - "sonar-reasoning-pro"
          - "sonar-reasoning"
          - "sonar-pro"
          - "sonar"
        fetch: false
      titleConvo: true
      titleModel: "current_model"
      forcePrompt: false
      titleMessageRole: "user"
      modelDisplayLabel: "Perplexity"
      dropParams: ["stop", "user", "frequency_penalty", "presence_penalty"]
fileConfig:
  endpoints:
    default:
      fileLimit: 5
      fileSizeLimit: 1024
      totalSizeLimit: 10240
      supportedMimeTypes:
        - "image/.*"
        - "text/.*"
        - "application/.*"
        - "application/x-sh"
    custom:
      fileLimit: 2
      fileSizeLimit: 1024  # Maximum size for an individual file in MB
      totalSizeLimit: 10240  # Maximum total size for all files in a single request in MB
      supportedMimeTypes:
        - "text/.*"
        - "application/.*"
        - "application/x-sh"

[youvnor]

thank you Danny.
Reading the documentation i totally did non understand this.

i can write something for the docs if you want.

[danny-avila]

The best place to start for custom endpoints is the following, I think it's pretty clear:
https://www.librechat.ai/docs/quick_start/custom_endpoints

Contributions to the docs are always welcome though. Feel free to make a PR:
https://github.com/LibreChat-AI/librechat.ai