Can Cursor generate videos? Images yes, video via MCP

Cursor's Agent has a built-in image generation tool but no video tool. To make video from Cursor, connect an MCP server that has one.

5 min readSume
All posts

Not on its own. Cursor's Agent can generate images with a built-in Image generation tool, but the tools listed in Cursor's docs include no video generation tool. To get a video from Cursor, connect an MCP server that has a video tool; the Agent then calls that tool and hands you a link to the clip.

Cursor's side comes from its Agent tools and MCP docs. The example video server is Sume's hosted MCP server, from the MCP quickstart and MCP tools and gates. All were read on 2026-09-28. Sume has no official Cursor integration: this is a custom MCP server entry.

Can Cursor generate images?

Yes. Cursor's docs describe the Image generation tool as generating images from text descriptions or reference images, for things like UI mockups, product assets, and architecture diagrams. Images are saved to your project's assets/ folder by default and shown inline in chat. The same page lists search, web, rules, file reading and editing, shell commands, a browser, and questions. None of them renders video.

From Cursor's Agent tools and MCP docs and Sume's MCP overview, read 2026-09-28.
You ask Cursor forBuilt in?How it happens
An imageYesImage generation tool; saved to assets/ by default
An image from an MCP serverVia MCPA tool can return a base64 image; Cursor attaches it to the chat
A video clipNoAn MCP server's video tool, such as Sume's generate_video

How do I make videos from Cursor?

Add a remote MCP server that has a video tool. Cursor reads servers from .cursor/mcp.json in your project, or from ~/.cursor/mcp.json for every project. For Sume's server that is one url entry plus an OAuth sign-in with Write turned on, since the default read-only grant can't call paid tools; Connect Claude Code, Cursor, or Codex to Sume has the entry.

Cursor asks for approval before using MCP tools by default, and you can expand a call to see its arguments, so each paid request is visible before you approve it.

What happens when the Agent asks for a video?

It gets a job, not a file. Sume's docs say video generation typically takes 30 seconds to several minutes, so the Agent calls generate_video, waits in slices of at most 55 seconds with jobs_wait (never resubmitting the paid create), then reads jobs_result. Waiting out long video jobs covers that loop.

The finished clip is a link: Sume's docs tell agents to prefer media.sume.com URLs in their reports and not to paste signed URLs into chat. The Agent can run shell commands, so ask it to download the clip if you want the file in your repo.

Can Cursor turn my image into a video?

Only if the image is reachable online. Sume's docs say hosted MCP cannot read files from your laptop, and its video docs say reference images must be accessible over public HTTPS. A file in your repo's assets/ folder is on your disk, so it needs a public HTTPS URL before a remote video tool can use it. How files reach an MCP server explains why.

What does it cost?

That depends on the video server you connect. On Sume, generate_video is a paid tool billed to the workspace wallet, and dry_run=true previews the cost without submitting; the cost section of the setup guide and API pricing have the details. Sume's basics page says hosted MCP still works but is not the primary path today, so a product that generates video calls the Developer API.

Sources

Related posts

More in Agents

All Agents posts

Written by Sume