Awesome MCP › Other MCP Servers

OpenDataMCP/OpenDataMCP

⭐ 154 Python repository created 2024-11-30

Open Data MCP is a Python project that connects public open data sources to LLM applications through the Model Context Protocol. It has two sides. On the access side it ships a CLI, published on PyPI as odmcp and normally run with uvx, that lists the available data providers, prints information about one, and writes or removes the matching MCP server entry in the Claude Desktop configuration. After a restart the client shows the new tools, so a user can ask questions answered from live data; the documented example is the ch_sbb provider, which serves disruption data for the Swiss SBB train network collected from data.sbb.ch. On the publishing side the repository is a place to contribute new datasets: each source becomes its own Python module under src/odmcp/providers, named by country code and organisation, and starts from a template file. Contributors decide whether a dataset is better served as an MCP tool, when it needs active querying, parameter filtering or computation, or as an MCP resource, when the content is static, small, file based or acts as reference documentation. Every tool or resource is expected to carry a purpose description, input and output schemas defined with Pydantic models, error handling and docstrings, plus tests covering basic behaviour, edge cases and failures. An experimental client script is included for checking a server before it is submitted. The project uses uv for environment and dependency management and pre-commit hooks for code quality, and states goals of minimal dependencies, type safety and standardisation so that many providers can be maintained together. It targets two audiences: users of MCP clients who want public datasets available inside a chat, and data publishers who want a distribution channel that makes their datasets discoverable by LLM tools.

https://github.com/OpenDataMCP/OpenDataMCP

mcpopen-datapythoncliclaude-desktopdatasetsuv