Cursor can now show your agent's context usage as an interactive report in a canvas. The context explorer breaks down where tokens go across the system prompt, tool definitions, rules, skills, and more.

Design Mode is now available in canvases. Select and annotate UI elements directly in a canvas to guide Cursor's edits and iterate more quickly.

With canvases, Cursor can create apps like dashboards, reports, and internal tools. Now you can publish a canvas and share it with your team via URL.

Learn more:

This kind of visibility is underrated. For long agent workflows, users need to know where context and tokens are going so they can steer the work before it drifts.

Context explorer is the unsexy feature that probably saves the most pain. Seeing tokens split across system prompt, tools, rules, and skills is like opening the hood and finding out your agent spent half its brain reading old house rules. Very useful.

This was very needed

This is useful because it turns context into something a team can review. Once rules, skills, and tool definitions have a cost centre, prompt debt stops being a vibe and starts showing up in review.

have you guys removed the context usage interactive canvas already?

Making the context window observable is underrated. Most agent failures I see come from invisible prompt bloat. If you can see the tokens, you can actually debug the agent instead of just rerolling.

I was excited about this one but team memberships only.

This is super useful, but I'm not able to see the View Report button... 🥲

@cursor_ai This is useful! I’m using version 3.7.12, but I’m still not able to see this option.

seeing where tokens go is genuinely useful 👀 the breakdown i keep wishing for is which ref got dropped when it compacted. that's usually the one holding the constraint i needed. does the explorer flag what fell out?

Good timing, @cursor_ai !

Context usage reports are underrated. Builders need to see which rules, tools, and files are eating the budget before they can make agents cheaper and more predictable. The next useful layer is a diff: what changed in context between a good run and a failed one.

Really needed

That's really cool

three big releases!