I built this because I work with AI agents a lot (for both coding and general research), and I found it's pretty annoying whenever after doing some research myself I still need to manually give my LLMs all the context so that they understand the constraints correctly / don't do something generic.

I'm also curious if we're going to live in a world where instead of relying on products (that are supposed to be commercially viable somehow, like what happened to Rewind), we can just build composable & hyper-personalized tools locally.