InfraNode with Python: the infranode package

Tutorial: fetch open data for German cities with the keyless Python package infranode, sync and async, plus ready-made LangChain and LlamaIndex tools for AI agents.

Install

pip install infranode

First request

One instance, then any endpoint via a method or the generic record(slug, endpoint). The data lives under payload, the source under attribution.

from infranode import InfraNode

api = InfraNode()
rec = api.weather("berlin")
print(rec.payload["temperature_c"], "degC")
print("Source:", rec.attribution["text"])   # always show attribution

# Any endpoint by name
fuel = api.record("hamburg", "fuel-prices")
cities = api.cities()                          # all 84 cities

Every record carries its own attribution (source and license URL). The open-data licenses require attribution, so display it to your users.

Async

For many parallel requests or asyncio apps:

import asyncio
from infranode import AsyncInfraNode

async def main():
    async with AsyncInfraNode() as api:
        rec = await api.air("muenchen")
        print(rec.payload)

asyncio.run(main())

Mind the rate limit of 300 requests per minute per IP; back off on 429.

As an AI agent tool: LangChain

pip install "infranode[langchain]"

from infranode.integrations.langchain import infranode_tools

tools = infranode_tools()   # bind these to your LangChain agent

As an AI agent tool: LlamaIndex

pip install "infranode[llamaindex]"

from infranode.integrations.llamaindex import infranode_tools

tools = infranode_tools()   # pass to your LlamaIndex agent

Both expose infranode_get_city_data(city, dataset) and infranode_list_cities(). The tool description lists valid datasets so the model picks the right one. If your runtime speaks MCP, use the InfraNode MCP server.

Learn more

Source and issues: github.com/street1983nk/infranode-python. For plain notebook analysis without the package, see City data with Python and pandas.

FAQ

Do I need an API key?

No. The package talks to the keyless, free InfraNode API. No key, no signup.

Sync or async?

Both. InfraNode is the synchronous client, AsyncInfraNode the asynchronous one. They share the same method API.

How is this different from the MCP server?

The MCP server is ideal when your runtime speaks MCP (Claude, Cursor). The Python package is the lightweight, MCP-free path for scripts, notebooks and LangChain/LlamaIndex agents.