Using Onnxruntime Custom Execution Provider for ORT-GenAI

In Onnxruntime, we are able to create custom Execution Providers via either shared provider or Plugin EP API.
Is there a way to integrate such custom EPs to onnxruntime-genai?
I only see CPU, QNN and CUDA and now WebGPU EP for onnxruntime-genai.

How can we integrate custom execution providers built via onnxruntime to onnxruntime-genai, since ORT-GenAI does not seem to recognize standalone ORT EPs. Am i missing something?

Start by reviewing how ORT-GenAI currently exposes CPU, QNN, CUDA, and WebGPU execution providers, then compare that with the shared provider and Plugin EP APIs mentioned in the issue. Determine whether standalone ONNX Runtime custom providers are intended to be supported and document the integration path or the missing design needed to make them available.

Written by the indexing model from the issue text.