Technical
Documentation

Set up Model Context Protocol (MCP) servers and find every global shortcut.

Extensibility

Model Context Protocol (MCP)

Vision implements the open Model Context Protocol via the official rmcp crate, allowing both OpenAI Responses and Claude Code to invoke custom local and remote tools.

~/Library/Application Support/com.vision.app/settings.jsonJSON
{
  "model": "gpt-5.5",
  "reasoning_effort": "low",
  "mcp_servers": {
    "filesystem": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/Users/yourname/projects"]
    },
    "postgres": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-postgres", "postgresql://localhost/mydb"]
    },
    "github": {
      "command": "docker",
      "args": ["run", "-i", "--rm", "-e", "GITHUB_PERSONAL_ACCESS_TOKEN", "mcp/github"]
    }
  }
}
✦

MCP servers can be added either via editing settings.json or directly from Vision's Dashboard window under the Integrations tab.

Keystrokes

Master shortcuts cheat sheet

All global shortcuts are fully customizable in Dashboard settings.

Global ActionDefault Key CombinationBehavior & Scope
Open Vision⌥SpaceSearch apps, Google, emoji and clipboard, or ask Vision.
Write with Vision⌥⇧SpaceWrite or rewrite text in the app you're using.
Files circle⌥⇧FConvert or share the files selected in Finder.
Vision Terminal⌥⇧TYour shell in tabs, with commands from plain English.
Vision Mode⌥⇧VVision full screen: talk hands-free, see everything at a glance.
Sound mixer⌥⇧MVolume for each app, plus your speakers and headphones.

Window-Specific Keys:

Space
Talk, in Vision Mode
Esc
Close Vision Mode or the Vision bar
⌘K
Clear the terminal and its scrollback
⌘Click
Open a link in the terminal
ClapClap
Wake Vision (when Clap to talk is on)

Summon Vision anywhere with ⌥ + Space on your Mac.

Apple Silicon & Intel · macOS 12+