we've gotten so much from this sub (feedback, bug reports, ideas) that it felt like the right place to call for a little community competition. if you build an AI SRE tool, put it on the leaderboard and let's see what you've got.
leaderboard: https://www.sregym.com/leaderboard
sregym breaks a real app on kubernetes and checks whether an AI agent can find the root cause and fix it. the sregym team verifies submitted results. to submit, open an issue with the submission label on github.com/SREGym/SREGym.
right now radar + claude code is in first place. five setups solved all 17 failures, radar was the fastest of them: 37s to diagnose and 66s to fix, vs 76s and 141s for devin oncall on the same model, with less than half the tokens.
taking first place felt like a good excuse to launch radar's AI investigations on product hunt today, so if you check it out we'd really appreciate your comments and input there too.
so: bring it on.…