Screenify Studio

Screenify Studio — MCP server

Every other screen-related MCP server lets an AI look at your screen. This one lets it make a video.

Ask Claude, Claude Code or Cursor for a demo and get back a finished file — recorded, styled and exported. No terminal, no flags, no editor.

"Record a 12-second demo of my landing page and put it on a café desk mockup."

Everything runs locally on your Mac. Nothing is uploaded.

https://github.com/user-attachments/assets/8c753dcb-0e53-4c00-acd6-0cd934b19415

_{Asked in a chat window; recorded, styled and exported by Screenify. No editor was opened.}

What it can do

Connect it to an AI assistant and ask for the finished video instead of driving the editor yourself:

• record a screen, window, area, simulator or a web page
• style the export — wallpapers, device mockups, photo mockups, auto-zoom, cinematic effects
• add camera moves and effects to the moments that matter
• add animated titles — and keep the ones you made in the app when exporting
• open a video or image you already have and style that instead
• export to mp4, mov, gif or webp, with platform presets

Everything runs locally on your Mac. Nothing is uploaded.

Install

Requires Screenify Studio 2.2.0 or newer. The bundle carries no binary — it launches the CLI that ships inside the app.

Claude Code and Cowork — plugin

This repository is also a Claude plugin. It bundles the MCP server with a skill that teaches Claude which Screenify flow to use for a request. In Claude Code:

claude --plugin-dir /path/to/screenify-mcp

The plugin's server runs locally, so it works in Claude Code and in Cowork sessions on your Mac. In a regular Claude chat, use the one-click bundle below instead.

What the plugin runs. scripts/screenify-mcp.sh looks for the screenify command that ships inside the Screenify Studio app (/Applications/Screenify Studio.app, or the copy the app installs at ~/.screenify/bin) and starts it as an MCP server over stdio. It downloads and installs nothing. If the app is missing, it returns an error that tells you where to get it.

Claude Desktop — one click

Download screenify.mcpb and open it. Claude Desktop installs the connector.

Any other MCP client

Open Screenify Studio → Settings → AI Assistants → Connect AI assistants. It registers the server with every AI client it finds on your Mac. Or from a terminal:

screenify mcp install

From a registry

Screenify is listed in the official MCP registry as studio.screenify/screenify, so clients that read the registry can find and install it without any of the steps above.

Docs

• MCP server guide
• CLI docs

Privacy Policy

Full policy: www.screenify.studio/privacy

In short, for this MCP server specifically:

• What it collects — nothing. The server runs locally on your Mac and exposes tools to an MCP client you already trust. It has no telemetry of its own.
• Where your data goes — recordings, projects and exported videos stay on your Mac, in your recordings folder and wherever you ask for the output. Nothing is uploaded by the server.
• Third parties — none for the recording and export tools. Two features reach out only when you explicitly use them: web recording fetches the URL you ask for, and the optional AI-director effects call the AI provider whose API key you supplied.
• Retention — the server keeps nothing after a job finishes; the files it produced are yours to keep or delete.
• Contact — [email protected]

Support

Issues with the MCP server or the app: [email protected]

License

This repository — the Claude plugin (a manifest, a launcher script and a skill), the MCP bundle (screenify.mcpb: a manifest, a launcher script and an icon) and its documentation — is MIT licensed.

Screenify Studio itself is proprietary software and is not covered by that license. The app provides the MCP server, the recording pipeline and the export engine; this repo only ships the small bundle that launches it.