InfraNode
Keyless remote MCP server for German public-infrastructure open data: weather, air quality, traffic, public transit, parking and roadworks across 84+ German cities (DWD, Umweltbundesamt, Mobilithek, GovData). 38 read-only tools.
README
InfraNode
A keyless, remote MCP server for German public-infrastructure open data.
German cities publish a lot of open data, but every source has its own format, fields and quirks, and several need portal registration. InfraNode normalizes ~20 categories, weather, air quality, public transit (incl. realtime departures), traffic, parking, charging, water levels, demographics, energy and more, for 84+ German cities behind one interface, and serves it as an MCP server. 38 read-only tools, no API key, no account, no install.
Sources include the Deutscher Wetterdienst (DWD), Umweltbundesamt (UBA), Mobilithek/DELFI, GovData, OpenStreetMap, Bundesnetzagentur, KBA, DIVI and more.
Connect
One line with Claude Code:
claude mcp add --transport http infranode https://mcp.infranode.dev/mcp
Any other MCP client, point it at the remote endpoint (Streamable HTTP):
{
"mcpServers": {
"infranode": { "url": "https://mcp.infranode.dev/mcp" }
}
}
- Cursor / Windsurf: add the block above to
~/.cursor/mcp.json(or the app's MCP settings). - VS Code:
code --add-mcp '{"name":"infranode","url":"https://mcp.infranode.dev/mcp"}' - Claude Desktop: add the same
mcpServersblock to yourclaude_desktop_config.json. - ChatGPT: add a connector with the URL
https://mcp.infranode.dev/mcp.
Full install guide, the complete tool manifest with example outputs, the permission model and an example transcript are in docs/mcp-install.md. The registry manifest is server.json.
Example
Ask your agent something like "What's the weather in Cologne right now?" and it
calls weather:
// weather(slug="koeln")
{
"data": {
"city_slug": "koeln",
"observed_at": "2026-06-18T13:00:00Z",
"source": "dwd",
"attribution": { "text": "Datenbasis: Deutscher Wetterdienst", "modified": true },
"payload": { "kind": "weather", "temperature_c": 30.4, "humidity": 43.0, "station_id": "02667" }
},
"meta": { "source_status": "ok", "cache_status": "hit", "correlation_id": "..." }
}
Every response follows the same { data, meta } envelope: each record carries
its attribution (license + source), and meta.source_status tells you whether
the upstream source delivered data, so a dead source degrades gracefully instead
of failing the call.
Tip: call
list_citiesfirst to discover valid city slugs (e.g.koeln,berlin,hamburg), then call any city-scoped tool.
Tools (38)
| Group | Tools |
|---|---|
| Discovery | list_cities, sources, compare (one resource across many cities) |
| Weather & environment | weather, weather_warnings, air_quality, air_quality_live, pollen_uv, water_level, flood |
| Mobility | transit, transit_departures, station_departures, station_arrivals, traffic, road_events, webcams, charging, sharing, fuel_prices |
| City & people | get_city, geo, demographics, indicators, unemployment, tourism, construction, accidents, health, icu_live, holidays, election, events, pois |
| Energy & vehicles | power_load, power_price, energy, vehicle_registrations |
All tools are annotated readOnlyHint: true / destructiveHint: false /
idempotentHint: true, so MCP clients can safely auto-allow them.
Prompts & resources
- Prompts:
city_briefing,compare_air_quality,commute_check, ready-made flows that chain several tools. - Resources:
infranode://citiesandinfranode://sources, browse the coverage catalog without a tool call.
How it behaves
- Keyless & read-only. No credentials, no writes, no user accounts.
- Canonical envelope.
{ data, meta }with per-source status and attribution. - Graceful degradation. A failing upstream returns
source_status, not an error. - Safe by design. SSRF and injection gates validate every request; inputs are checked against fixed allowlists.
See SECURITY.md for the security model.
Self-host (optional)
You don't need to, the hosted endpoint above is the fastest path. But the code is open. Run the API stack locally with Docker (Compose v2):
cp .env.example .env # example config, contains NO real secrets
docker compose -f deploy/docker-compose.yml up
curl http://localhost/api/v1/health # -> {"status":"ok","version":"1.0.0","redis":true}
To run the MCP server itself locally over stdio (against the public API):
uv sync --group mcp
INFRANODE_MCP_API_BASE=https://infranode.dev/api/v1 uv run python -m infranode.mcp.server
All settings use the INFRANODE_ env prefix (see .env.example); each data
source is toggled by its own INFRANODE_ENABLE_* flag. Real secrets are never
committed, only .env.example is versioned and CI runs a gitleaks scan.
License: code and data are separate
- Code: Apache-2.0 (see LICENSE).
- Data: the open data served through InfraNode keeps the licenses of its
upstream sources (e.g. ODbL for OpenStreetMap, DL-DE-BY for GovData, attribution
for DWD). These data licenses and attribution are tracked separately in
DATA-LICENSES.md. The Apache-2.0 license covers only the API source code, not the passed-through data.
Contributing
Contributions are welcome. Setup, gate commands and the secret rule are in CONTRIBUTING.md.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.