Just a heads up, I’m playing with agentic verification of the entire OpenHands application, based on a feature map documenting expected behavior, using the UI as a user would. I’d love early feedback! feat: add weekly feature verification skills and map The UI Audit that the agent managed to do on the PR above ^^ shows, if it was necessary, that LLM capabilities are here, they are definitely here, and we can use them to… stop worrying about bugs on verifiable core user-facing behaviors.