I had hopes for the “shadow workspace” which I saw in the docs, but this setting doesn’t exist for me. Letting the model know about what error was introduced by the change - will require one more call to the LLM so at this point it seems we’re getting into a feature request of running many processes and agents behind the scenes , will probably cost us more, but I’m ready to pay. .
The closest option is to hover over the error and use the “AI Fix in Chat” button, but that will open a new chat. It would be useful to have an @Errors symbol in the chat or to have the errors automatically provided in the chat context.
So re: Screenshot 01, how can we have Add to Chat (will add to any existing chat) just like in Screenshot 03?
With the option for:
(in the case of more than 1 problem: “Add all @file problems to current chat”)
or individually for each error
The important part for me is the context shift, so including the fix in the current workflow - rather than separate chat window - is key. It also means subsequent responses in that chat context remember the fix, and don’t attempt to reintroduce the original error.
Thanks for the response @litecode and I think to remain within the narrow focus of this issue the base requirements would be:
Where the Problems tab identifies 1 or more problems in a file:
allow the user to click/right-click on the filename and select, “Add to Chat” (assuming this adds all file errors with @file context to the filename to the current chat)
allow the user to click/right-click on each individual error and select, “Add to Chat” (assuming this adds this specific error, with @file context to the filename to the current chat)
Which I think is also inline with @FARZ feature request.
So it is, let’s see how it helps or not. I realise it may mean it marks its own homework before replying but I think in context to this feature it still won’t have context to the errors it’s introducing.