What features are planned for the future management of MCP servers?

[wrpromail]

1. For example, after integrating a new MCP service instance, can its health status be monitored?
2. Can the system act as a proxy for MCP servers—similar to how it proxies language models—and provide audit information such as logs?
3. Currently, I can view logs by invoking a tool, inputting the name of the MCP server in the request, which then calls a specific model. However, this requires the caller to know which MCP servers are configured in litellm. For external users, they may not—and should not—be aware of the internal MCP services we manage. I’d also prefer not to expose these implementation details. Is there a way to hide the internal configuration while still bundling MCP servers with models?