Feature request: image support

Hello,
I haven't found any way how to paste the image (or its path like in a native pi agent), so that it could be processed by the model. Are you planning to add this functionality into your pi wrapper too? Thanks!