New in v15.6.0 - ✨ Vision

[olimorris]

Vision.mp4

Hi All,

I'm very happy to announce support for vision/image inputs in v15.6.0 for Copilot, OpenAI, Anthropic, Gemini and Mistral (if I've missed any adapters that do support it, please make a PR).

How it works

There are two ways of getting images into CodeCompanion:

1. Via the image slash command
2. Via the img-clip.nvim plugin

If you're on Mac, you may have some challenges with img-clip.nvim currently as it uses pngpaste which seems to be broken.

Final Notes

I've added direct support for image pickers to the snacks and default providers. PRs of course welcome to add support for the other providers.

Enjoy!

[bassamsdata]

Ah, the long-awaited feature, this is fantastic! Thank you for adding it and it’s working perfectly. Just one quick question: which picker are you using to display the image preview?

[bassamsdata]

Ah, it’s probably snacks.nvim. I guess it might be time for me to switch from mini.pick to snacks too 🙂

[ravitemer]

Just pointing it out that with v5.3.0 MCP Hub also supports images with codecompanion. See more details here ravitemer/mcphub.nvim#136

cc-image-gen.mp4

[aligit]

Hey @olimorris,
Thank you for turning my suggestion into a revolutionary feature. I'm really thankful of having what started from discussions. Thank you Thank you sir

[Jufralice]

Hello,
Thank you for this amazing work! Having the possibility to add images in our context is a game changer.

Is it plan to add the 'image' data-type to the workspace file?
I for example have a workspace-group that include a .md file with the documentation of the implementation of a new feature in my project and I also draw the data-model schema changes in excalidraw. Having the possibility to include that excalidraw .png export to my workspace group would be cool.

Thanks.