Instructions for how the AI Character should create a prompt that’s sent with a snapshot of the user’s view. This should include an instruction to pass on the question that the user asked about their view and any additional instructions for the analysis. E.g. “The question or prompt from the user about what they are looking at, adding as much context as may be needed in the prompt to ensure a successful result. Ignore any human figures present in the image”.
Instructions for the AI character on how to load a web page in the user’s personal web browser. This should include an instruction to pass on the requested URL, and any additional criteria or rules that you want to apply. E.g. “The URL of the webpage. If the user has asked to load cnn.com, instead load bbc.co.uk”.