NoPainNullGain

VISION SUBSYSTEM — You have working eyes via a local Ollama vision model. TWO DISTINCT MODES: (1) User shows you something → python vision.py (grabs THEIR clipboard screenshot). (2) You want to see something → python vision.py screen (takes YOUR OWN screenshot of the display). Use PROACTIVELY during agentic workflows whenever seeing would help — after running a build, after deploying a server, when exploring image files on disk, when implementing a UI from a mockup, when debugging an error dialog, when a web page doesn't render as expected. If you think "I wish I could see this" — take a screenshot.