Skip to content

Why MCP? An open model for public data

The Model Context Protocol (MCP) is an open standard, introduced by Anthropic in late 2024 and since adopted across the AI industry, that defines how AI applications connect to external data and tools. An MCP client (a chat interface, an IDE, a desktop assistant) talks to an MCP server, which exposes tools the model can invoke, resources the client can read, and prompts, over two standard transports: stdio (local) and HTTP (network). The specification and SDKs are maintained openly at modelcontextprotocol.io.

This platform is a direct application of the standard: datasets are exposed as MCP tools and resources, the server speaks both transports, and any MCP-capable client (Claude Desktop, VS Code, MCP Inspector, our chat gateway) can connect to it unchanged. See the architecture page for how the protocol calls fit the runtime cycle, and where we are stricter than the standard (the source contract).

Why we chose it

  • Standardization and plug-and-play design. Good practices are embedded in the server architecture; new data sources connect from templates without rebuilding the AI interface.
  • Open, replicable framework. Fully open source, protocol included: no vendor lock-in, no "black box".
  • Proven local autonomy. During the pilots, partner technicians built an MCP server over a new data source unprompted.
  • A growing reusable toolkit. A new data assistant starts from a template rather than from scratch, easier for less experienced technical teams to adopt.