Currently struggling while trying to use the tool, every new instruction makes the previous verse script worse, adding bugs and compilation errors.
Assistant can deliver verse code that doesn’t compile. Would love to have a way to send the compilation errors in bulk back to the assistant so they know what they did wrong.
Also the assistant doesn’t fully know the devices and cant test whether their suggestions are actionable. Would be nice to be able to test the solutions proposed before hand to check if doable within the engine.
Also, having the ability to write more than 4k lines as a reply will allow us to paste correct scripts that we want the AI to compare with the wrong versions and fix its mistakes. This limitation currently is preventing a lot of discussions with the assistant.