Integrum - Reflection based MCP server from any Python Module/Library [P]

Hi! I wrote a small library called Integrum that allows you to quickly create MCP server(s) from existing Python libraries or modules. It has a CLI so it is really easy to use (I hope!).

It is open-source (MIT) and on PyPI, so sharing it here if somebody finds it useful.

Code: https://github.com/alphadeepai/integrum

Blog post: https://nmilosev.svbtle.com/integrum-reflection-based-mcp-server-from-any-python-module-or-library

As an example I gave Gemma 4 access to scikit-learn and told it to create a RF classifier for a toy dataset (Iris) and it did it rather well! Here is that example session..

Obviously an interesting discussion is this approach vs. just letting the agents write code: I think this approach is more formal and therefore easier to verify.

I haven't found any other reflection based approaches like this, let me know if I missed some.