Tinybird MCP

Tinybird is a real-time analytics backend where teams stream event data into managed data sources, shape it with SQL pipes, and publish those pipes as parameterized HTTP endpoints. The Tinybird MCP server puts that workspace behind the Model Context Protocol, so an assistant can inspect the schema, run queries, and read results without a developer opening the CLI or hand-writing SQL against an unfamiliar dataset. Once connected with a workspace token, the server lets an agent enumerate the data sources in a workspace and read their column names and types, list the pipes and API endpoints that have been published along with the parameters each one accepts, execute read-only SQL against workspace data, and call a published endpoint with arguments to get the same result an application would. Service data sources that record ingestion and request activity are reachable the same way, which lets an agent answer questions about pipeline health rather than only about business metrics. Token scope determines what is visible, so access can be narrowed to a subset of the workspace. It suits data and analytics engineers who want a conversational way to explore a Tinybird workspace, and product teams building agents that need to answer questions from live event data instead of a static export. Because the query surface is SQL over defined schemas, answers stay tied to the actual tables rather than to a summary the model has memorized.