Re-use / Usage agnostic implementation of @Tool?

[0xSpell]

I am currently researching ways of embedding LLM agents in my stack. While MCP sounded right at first, it seems that currently remote MCP servers are kind of difficult.
Fortunately, the big players seem to also support receiving Tool definitions in API requests to trigger additional queries to the backend.

It might probably a bit out of scope for this module, but maybe the Tool definition generation together with schemas etc could be re-used so that its possible to use tool definitions for the MCP server as well as using it for API calls?

I will most likely be experimenting a bit in that direction anyways.

Kind regards,
Markus

[0xSpell]

I'm sorry, I just noticed that the langchain4j project basically does that and I'm a bit wrong here.