I created a QA-by-openhands workflow so OpenHands can help us QA each PR by actually running code... I accidentally created a perfect little recursion joke 😅 This SDK PR adds a workflow that launches the QA agent. The QA agent then goes to verify the PR by monitoring that exact same QA agent workflow run. So the workflow starts the agent, and the agent patiently watches the workflow, which cannot finish because the agent is still there watching it. Eventually the agent loses patience and says the workflow is taking too long. Which is amazing, because the thing taking too long was basically the agent itself 😅