VS Code 1.112 Can Now Let Agents Read Images

VS Code 1.112 adds native image analysis for agents, which means screenshots and other image files can now become actual inputs instead of things you have to manually describe.

This hands-on test uses three Microsoft Foundry leaderboard screenshots to see whether an agent can turn chart-heavy visuals into something useful for developers. The result is pretty practical: less transcription, more structured extraction, with the usual reminder that humans still need to sanity-check the output.

If you work with screenshots, dashboards, charts, or UI mockups, this one is worth a look.

Full article here: Hands On with VS Code 1.112’s New Image Analysis for Agents - Visual Studio Magazine

What would you want an agent to do first with image input?