This tool maps over a set of documents with a question, and reduces them to an answer.
This is a sort of brute-force LLM-based tool, not a sophisticated RAG/MCP system. Naive, one-shot inference has a role, however, in generating the kinds of meta-context that can improve RAG/MCP performance.
If you like this sort of thing, check out the infinitely more mature and useful llm tool.
Frankly, this repo should have been a plugin for the above utility.
$ ad-rem ./recipes/ "How's my diet?"