Full featured Dataverse MCP Server

Dataverse already offers an MCP server implementation (https://github.com/gdcc/mcp-dataverse) that enables search and access to dataset metadata using the Croissant ML format. In addition, there are ongoing efforts to extend pyDataverse with utilities that simplify the creation of custom MCP servers built on top of the Dataverse REST API.

Our goal is to take this one step further by developing a fully automated, complete MCP integration for Dataverse. Instead of manually mapping endpoints or updating the server when Dataverse evolves, our approach will expose all Dataverse API functionality through MCP dynamically. As new API endpoints are added to Dataverse, they would become accessible via MCP automatically without requiring code changes or maintenance.

To address the challenge of LLMs interacting with MCP instances that expose a very large number of tools, we also propose the concept of MCP profiles. Profiles would allow administrators or users to define which subsets of the Dataverse API/functionality should be surfaced through MCP, making the interface more efficient, safer to use, and tailored to specific workflows.