Comment by stillsut
10 days ago
Working on a script to let people launch claude code, gemini cli, and aider agents on the same task and compare results. Handles all the git worktree stuff for you so they can run in parallel. Kind of like OpenRouter but for CLI agents.
What feels really cool is using the framework to build framework, I've got 150 ai generated commits stacked on top of each other, you can see every prompt and accepted solution in the dev logs I produce: https://github.com/sutt/agro/blob/master/docs/dev-summary-v1...
you may want to check out https://github.com/Forward-Operators/prr It's abandoned now as the company is gone, but back then we were trying to find an easy way to compare LLM outputs