As a suggestion, I recommend using claude-3.5-sonnet. It is already pointed to the updated model from October. The 20241022 version may therefore produce strange results.
I think I did this myself sometimes. One example would be a solution that did not work entirely then I would for example post an image of javascript traceback with relevant parts of the page as answer.
All context is already incorporated in the image and then there is the previous context of the chat. Just adding “does not work”-text to the image does not really add context there.
Worked fine for these occasions.
Just to clarify, this was never the first message, but a follow-up to the previous conversation.
sometimes the cursor asks me to provide the error that appears in the browser console, and usually there are so many errors, it’s faster to take a screenshot and send it to chat, instead of having to type, “here’s a screenshot of the error in the browser console”. Or also sometimes the cursor asks for a screenshot of a table in the browser, when we send it there is no need to type “This is a screenshot”, right?