More powerful integration with AI features
- Ability to upload an image in the Gemini chat. I should be able to screenshot a figma design and have Gemini implement it as best as it can.
- Ability to reference local code. Similar to Cursor IDE, I should be able to tell Gemini: "Look at this @file as a reference", or: "Only use the components under @example/components"
- Ability to customize a default prompt. If I always use certain conventions I don't want to repeat them over and over. I should be able to write them once under .idx/prompt.txt and have Gemini use them for context.
Francesco Sardo
shared this idea
-
Joseph Hoff commented
I've had an extreme amount of aggrivation with Gemini as I have been kinda just teaching myself and not using all the other tools available but I've managed to learn to navigate with little thanks and a lot of swearing at Gemini for repeating steps that it had just told me and when I would say that didn't work he would say the same **** over and over and he would always cut me off as I'm speaking or he would just dip and be no responsive or not look at my screen... we'll my the hall am I sharing my screen then...yeah Gemini has been very aggravating... I had actually made it almost 3 time as far on the limited questions with the free version of chat gpt...oh and not being able to share info and having to constantly exit out of screen share sucks too....but I think one of the biggest thing would be to add the ability to copy and paste the text of info it tells us in live mode