The trend toward CLI-first tooling is interesting — devs increasingly want to stay in the terminal rather than context-switching to a browser.

Nice project. Ideally it should be possible to run arbitrary graph or datalog queries to get relevant items but a hierarchical organization w/ basic content type tagging + vector similarity search is a good starting point.