Thanks again for the contribution. After looking into it, there are two prerequisites before implementing the Colanode MCP:

1. Design the structure of a public API - which defines how we want to allow external parties modify content in Colanode and also how the authorization works (personal access tokens, API keys etc). This is different from the client/server flow that is currently being used by the web & desktop app, given the full sync & local-first approach.
2. Think about which tools are the most suitable for LLMs to provide better performance/quality when working with Colanode. We are still experimenting with AI in general and this is something that we are already looking into.

We might need to wait a bit before we fully implement the MCP server. However, this is a good start and I'm leaving this open. We will continue working from here.