`jdr` is a minimal "deep search" implementation that achieves state-of-the-art / near state-of-the-art performance on all of the benchmarks we've tested.
Most of the open-source "deep search" / "deep research" implementations we saw are pretty complicated and feature rich. Before starting this, we assumed you'd need lots of tricks to get good performance. The goal of `jdr` was to do "the simplest possible thing" and we were pleased & surprised to see how well it performs.
Mature frameworks are great, but sometimes it's nice to have an implementation where all of the code "fits in your head". Good for experiments, learning how this stuff works, or as a starting point for your own project.
