Five new Claude Code repos — autoresearch, OpenSpace, CLI-Anything, peers, gws

If you use Claude code then here are five open source projects that just came out this month that you got to check out. Now you've probably heard of this first one unless you are living under a rock and that is auto research by Carpathi. This basically brings in machine learning type experiments into Claude code. We just point Claude code at something we want to improve. Imagine you created some sort of Python app and you want it to run faster and then auto research will run a bunch of experiments continuously to try to improve that product. The example given here had 83 experiments run in succession and it kept 15 of the improvements. Now you need to make sure you're testing something that sort of has a binary yes that improved no it didn't like the speed of something but if you do that this is super useful. Number two on the list is open space and this is all about self-improving skills. You point this project at some sort of a task and then it evaluates your skill usage and put some into three buckets auto fix if the skill breaks auto improve if the skill is actually working but the patterns can be better and then auto learn. So essentially this captures winning workflows from actual usage from their tests on real world use cases. They're saying it improves token consumption by 45 percent and that it increased the income generated using the GDP Val economic benchmark by 4.2 times. Now the third one on the list is from the same authors and that is CLI anything and this…