Cursor Canvas Guide: What It Is and What to Build With It

Cursor's June 2026 Canvas update added Design ModeA way to point at an element in Cursor's built-in browser and change it directly, instead of describing it in words. Press Enter for the full definition. inside canvases and an interactive context usage report. It also added a Debug with Agent button, full-screen shared canvases and embedded canvas buttons that can run prompts.

Cursor's release notes and announcements say what changed. They don't say how a team should act on it. The steps below are our suggested order, and the next section lists what to watch out for along the way.

1. 1Use a canvas when the output is an artifact, dashboard, audit or decision report.
2. 2Ask for a context usage report when the agent feels slow, expensive or distracted.
3. 3Use Design ModeA way to point at an element in Cursor's built-in browser and change it directly, instead of describing it in words. Press Enter for the full definition. to fix canvas UX without describing every element in prose.
4. 4Share a canvas only after its labels, caveats and permissions fit the audience.

Each item below is either a fact about the release, such as its beta status or plan access, or our advice for using it safely. Plan around each one, and recheck the facts when Cursor ships an update.

• Don't hide caveats or source limits inside visuals. A caveat left off a shared canvas stays off for every reader until you republish.
• Give each embedded prompt button a clear purpose, because each one starts a workflow.
• Use shared canvases as evidence. They don't replace source files or PR review.

You can build an enriched view from a plain prompt. Open a canvas from the plus menu, and Cursor renders a polished visualization with its built-in UI frameworks. You don't have to ask for React or hand over brand guidelines. The agent takes an image, a CSV or your codebase as input and turns it into something you can look at.

Four conditions have to hold, and an account that isn't on a team fails at least one of them, whatever its plan.

Publish, on the canvas toolbar, uploads a live snapshot that teammates open in a browser, with the same layout, charts and tables. Shared canvases are available on Premium and higher, Teams, Enterprise, and legacy Pro, Pro+ and Ultra, which leaves out Free and Plus. Only members of your team can open a share, so you also have to be on a team, though a member on an individual paid plan who belongs to a team can share. Sharing also needs a privacy mode that allows data storage, so Legacy Privacy ModeCursor's setting that guarantees code data is not used for training by Cursor or its model providers, and that an admin can enforce org-wide; data-retention terms are a separate, contractual layer. Press Enter for the full definition. blocks it.

The privacy condition is the one that catches enterprises, because it is a privacy decision and the person clicking Publish rarely made it. Team admins can also turn shared canvases off for the whole organization from team settings. Check both before you promise someone a link.

The Shared Canvases page on the dashboard lists only the canvases you published yourself, so ask a teammate for the link to theirs.

Publishing again refreshes the existing share instead of creating a second one, so the link you pasted into a doc keeps working and shows the new version. That is why the source note and the date belong inside the canvas itself, where every refresh carries them.

Start over for anything bigger than a tweak. Cursor's docs say to revert and prompt again with more detail, because that is usually faster than nudging through small follow-ups. If the layout is wrong, tell Cursor what to change instead of editing it by hand, and keep manual source edits for small tweaks.

The docs give one more instruction, and we rank it first. If the numbers look stale or wrong, ask Cursor to rerun the underlying query or show its work. The tempting alternative is to fix the figure in the source. That changes what the canvas displays and leaves the query that produced it as wrong as before, ready for someone who trusts it to rerun later with fresh data.

The second time someone asks for the same shape. Cursor's docs list four parts for a canvas skill:

• A trigger description, so Cursor knows when to reach for the skill.
• Layout instructions that name the sections, stats and tables.
• The data sources or queries that populate the view.
• Formatting rules, such as units, date ranges and sort order.

Once the skill is in place, a short prompt regenerates the canvas with fresh data, and every teammate who uses the skill gets the same output shape.

For your own work, a skill is often more than you need. Each canvas is saved to your workspace's canvas list, where you can reopen and rerun it. Package a skill when other people need the same canvas.

Of the four parts, formatting rules pay off soonest. Two versions of the same dependency audit, one sorted by severity and one covering a different date range, look like a disagreement about the codebase when the only difference is the prompt. Pinning the units and the date window costs one line in the skill and ends that argument.

A skill won't fix the main limit of a canvas. A canvas is an artifact rendered next to the chat, not a file in the repo. If the output needs review, will be merged, or will have other code depending on it, it belongs in a pull request. A shared canvas is good evidence for a decision, and it doesn't replace the source or its review.