An MCP server that lets any MCP-compatible client (e.g. Claude Desktop) query, push and manage data in a Tinybird Workspace via Tinybird’s APIs.
https://github.com/tinybirdco/mcp-tinybirdSkip the dashboard dance. Query your real-time data sources, analyze trends, and push insights directly through Claude Desktop with natural language commands.
You know the drill: Claude helps you brainstorm an analysis approach, then you context-switch to Tinybird's UI to write SQL, export results, paste them back into Claude, and repeat. Every insight requires bouncing between tools, breaking your analytical flow.
This MCP server eliminates that friction entirely. Your Tinybird workspace becomes Claude's direct data source, turning conversations into executable analytics.
Direct Data Access: Query any data source in your Tinybird workspace without leaving Claude. Schema introspection means Claude understands your data structure and can suggest relevant queries.
Pipe Execution: Run existing API endpoints with dynamic parameters. That revenue-by-cohort analysis you built? Just ask Claude to run it for Q3 with different segmentation.
Real-time Iteration: Modify queries on the fly based on initial results. Claude can analyze patterns, suggest optimizations, and drill deeper into anomalies without manual SQL rewriting.
Performance Intelligence: Built-in query analysis using Tinybird's explain API. Claude can identify indexing opportunities and partition key optimizations while you're exploring data.
Insight Persistence: Save discoveries directly to your workspace as you uncover them, building an organizational knowledge base of data insights.
Bluesky Social Analysis: One developer analyzed Bluesky engagement patterns by asking Claude to explore user behavior data. No SQL writing, no CSV exports - just conversational data exploration that revealed viral content patterns.
Web Analytics Deep Dives: Instead of clicking through dashboard filters, ask Claude to "show me conversion rates by traffic source for mobile users in the last 30 days, then compare to desktop performance." It queries your web analytics starter kit data sources and provides comparative analysis.
Operational Monitoring: "What's causing the spike in API errors since 2 PM?" Claude queries your logs data source, identifies patterns, and suggests which services to investigate first.
Install via Smithery or mcp-get, add your Tinybird credentials to Claude Desktop's config, and restart. Your data sources appear as tools Claude can use immediately.
The server introspects your workspace on startup, understanding your data schemas and available pipes. Claude gains context about your business metrics without you explaining table structures or column meanings.
Custom prompts let you define analytical workflows specific to your domain. Set up a "retail analysis" prompt that guides Claude through your sales funnel data, or a "product metrics" workflow for feature adoption analysis.
Built on Python with the MCP SDK, supporting both stdio and SSE modes. The server handles Tinybird authentication, query optimization, and result formatting for LLM consumption.
Full Tinybird API integration means you can push data files, create new data sources, and even analyze query performance - all through conversational interfaces.
Production-ready with monitoring hooks and deployment flexibility. Run locally during development, then deploy to your infrastructure with the included Docker configuration.
Your analytical conversations become more productive when your data platform participates directly in them. No more context switching, no more manual query translation - just fluid exploration of your real-time data through natural language.