Powerskills

powerskills

Windows automation toolkit workflow for agents that need desktop control, Outlook access, Edge automation, or structured PowerShell-driven system actions.

Overview

Use this skill when:

• The task is Windows-specific and shell-only interaction is insufficient
• You need Outlook email or calendar access from the desktop client
• You need Edge browser control through CDP
• You need screenshots, window management, or keyboard automation
• You want PowerShell actions to return structured JSON for agent consumption

Prerequisites

• Windows environment
• PowerShell execution policy allows local script execution
• The required PowerSkills scripts and config are installed

Core Model

PowerSkills exposes several focused capabilities through a shared command surface:

• outlook for inbox and calendar actions
• browser for Edge automation
• desktop for screenshots, windows, and keystrokes
• system for processes, shell commands, and system inspection

Workflow

1. Confirm the task truly requires Windows-native automation rather than ordinary shell commands.
2. Choose the narrowest PowerSkills capability that fits.
3. Run the action through the PowerSkills entrypoint.
4. Parse the JSON result instead of scraping console text.
5. Verify the side effect with a follow-up read action when possible.

Response Shape

Prefer tools or wrappers that return a structured payload such as:

{
  "status": "success",
  "exit_code": 0,
  "data": {},
  "timestamp": "..."
}

Use Cases

• Pull the latest Outlook messages for a workflow
• Inspect or control a visible Windows application
• Capture screenshots during desktop troubleshooting
• Query processes, ports, or machine state in a structured way

Guardrails

• Use the smallest scope possible for desktop automation
• Verify user-facing actions before repeating them
• Prefer read-only inspection before mutation
• Keep environment-specific assumptions explicit, especially around Edge and Outlook availability