FlowMCP
FlowMCP v4.2 — Grading as a Versioned Standard · Read the announcement Schema-driven data access
Schema-driven data access
for AI agents.
FlowMCP is a schema format that standardises how AI agents reach any data source: write the connection once, reuse it everywhere. Each schema is normalised to one uniform interface — and kept separate from its grading and security tests — so hundreds of sources look alike to the model and stay cheap to maintain.
291 data sources · 1,608 AI-callable tools · AI-native, built for deep research.
Open source · MIT · GitHub-native · No vendor lock-in
Used at Berlin Mobility Hackathon 2026 · DB InfraGO
Connects to
- JSON
- XML
- SQLite
- GTFS
- OParl
- CSV