updated semantic, etc

tweaking clients (adding logging to them and to the aggregator function), making a domain folder for stuff like keywords, etc

replaced arxiv api test with client test, added academic search, etc

pubmed test works, getting 5 articles back, with title, author, abstract, doi (lotta N/a tho), and pubmed link (also N/As)

tweaky tweaks

forgot to remove the imports i wasnt using

forgot to remove the imports i wasnt using

decided to split api concerns, split pubmed off

updated chnages.md, fiddled with api clients, http.py

redid the venv and lock file, gradio pre-release messing with add/sync/ etc because you need to explicutly allow pre-release with any uv actions that re-reseolves

adding a tenacity based retry handler, etc

tidyer api stuff, etc

more api thoughts - static methods, reponse handling, etc

some api bits, still figuring out the best shape/design a little

archictecture and structure notes, etc

rearranging and a few bits of note taking

md edit, review doc

etc

etc

etc

some hf bits and bobs

moving stuff around, etc

memory system integration into corvus

memory system rathewr than cache reliance, so the group has shared context and a sorta 'lived memory' like a real gc

api client code, incl classes for pubmed, semantic, and arxiv clients

stickiing with cache as is for now, will refactor/redesign if i'm right

adding the corvus first draft (altho on reflection i think tha cache is poorly designed/not quite right as is, suspect it may dominate rather than support

structure work, adding some basic stuff, testing the arxiv made me realise i dont want to use that as main/1st search (lack of DOI, unhelpful structure of some results, etc)

doing some boiler and testing api stuff