Add policy enforcement for OS-level UI automation

[L1AD]

This server gives agents direct control over Windows UI:

• mouse_click -- clicks anywhere on screen
• keyboard_type -- types arbitrary text into any application
• launch_application -- opens executables

An agent with keyboard_type and mouse_click can interact with any application on the system -- including admin tools, terminals, and browsers with authenticated sessions.

Suggestion: Consider documenting support for PolicyLayer/Intercept, an open-source MCP proxy that enforces policies on tool calls before execution.

Example policy:

version: "1"
default: allow

tools:
  launch_application:
    rules:
      - action: require_approval

  keyboard_type:
    rules:
      - rate_limit: 30/minute

  mouse_click:
    rules:
      - rate_limit: 60/minute

One line to set up: npx -y @policylayer/intercept init