GitHub Copilot in VS Code: Image and Video Support in Copilot Chat

GitHub Copilot

GitHub Copilot's chat interface in VS Code now supports attaching screenshots and videos to messages, and agents can return images or recordings of changes for review in a built-in carousel with zoom, pan, and thumbnail navigation. This extends Copilot's multimodal capabilities from static image (vision) input to full video support, making it easier to share visual context β€” such as UI states, error screens, and recorded workflows β€” directly in the chat interface as part of the VS Code v1.111–v1.115 March releases.


Multimodal Chat: Images and Videos Now in GitHub Copilot for VS Code

GitHub Copilot in Visual Studio Code has expanded its multimodal capabilities in the March 2026 releases, adding support for both image and video attachments in Copilot Chat. This marks a step forward from earlier image-only vision support, giving developers richer ways to share context with the agent.

What's New

Developers can now attach screenshots and videos directly to chat messages sent to Copilot agents. This is particularly useful for:

  • Sharing UI states or visual bugs that are difficult to describe in text
  • Providing recorded workflows or reproduction steps for complex issues
  • Sending before/after screenshots of code output or rendered results

On the output side, agents can now return images or video recordings of the changes they have made, displayed in a carousel view with zoom, pan, and thumbnail navigation built in. This means developers do not need to switch to a separate window to inspect what the agent modified β€” the visual output is surfaced inline in the chat interface.

Context on Vision Support

GitHub Copilot's vision input (image attachments) first appeared in early 2025 in VS Code and Visual Studio. The March 2026 update extends this to full video support and improves the viewing experience for agent-returned visuals with the new inline carousel interface. The combination of video input and visual output review makes the feature substantially more useful for frontend and UI-heavy workflows.