DesktopVisionMCP. Show, don't tell.
Stop describing your screen or pasting screenshots to AI.
Start new thread

DesktopVisionMCP. Show, don't tell. - Stop describing your screen or pasting screenshots to AI.

by•
You describe your screen to your assistant. You screenshot it, crop it, paste it, and still leave out the thing that mattered. DesktopVisionMCP lets it look instead. A macOS menu bar app and a plain MCP server, so it works with Claude, ChatGPT, Codex, Gemini, or whatever you move to next. It captures only when asked. It never records, never clicks, never types, and its state is always in the menu bar. Captures go to the assistant that asked, and nowhere else.

Add a comment

Replies

Best
Maker
📌
Thanks for taking a look. A few things I expect to come up: - Privacy: DesktopVisionMCP captures only when your assistant asks, never continuously, and the capture goes to the assistant that asked and nowhere else. Nothing reaches me. Worth being straight about the other half: if your assistant runs in the cloud, then your screen reaches that provider, the same way a screenshot you paste would. The app shows its state in the menu bar at all times and stops completely when you stop it. - Other options (screenshot and paste, screen-sharing tools): the point here is that there is nothing new to live in. It is a plain MCP server, so it shows up inside Claude, ChatGPT, Codex, Gemini, or whatever you switch to next. You ask your actual question instead of narrating what is on screen, and the assistant sees the real stack trace, the real diff, the real spreadsheet. - It does not click, type, or act on your machine. It reads the screen and hands back an image. That is the whole surface. PS for PH: PRODUCTHUNT20 at checkout for 20% off, first 20 only. Happy to go deep on any of these.

 Upvoted, qq when the assistant triggers a screen capture, does macOS prompt for the standard system screen recording permissions dialog each time or just on initial setup?