WinWright

Name: WinWright
Author: civyk-official

@civyk-official

WinWright について

Playwright-style MCP server for Windows desktop, browser, and system automation. 110 tools for WPF, WinForms, Win32, Chrome/Edge via UIA3 and CDP. Record once with an AI agent, replay forever with deterministic JSON scripts. Self-healing selectors via winwright heal. Single self-

#windows #desktop-automation #browser-automation #testing #mcp #uia3 #cdp #playwright #rpa #winforms #wpf

基本情報

カテゴリ

ブラウザ自動化

トランスポート

stdio

公開者

civyk-official

投稿者

Veerabhadra Rao Ponna

設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "winwright": {
      "command": "C:/path/to/Civyk.WinWright.Mcp.exe",
      "args": [
        "mcp"
      ]
    }
  }
}

ツール

~12

—

概要

What is WinWright?

WinWright is a Windows automation server for the Model Context Protocol (MCP) that provides ~59 consolidated tools for desktop (WPF, WinForms, Win32), browser (Chrome/Edge via CDP), and system management. It is designed for QA engineers, developers building AI agents, teams extracting data from legacy apps with no API, and anyone automating repetitive multi-app workflows on Windows.

How to use WinWright?

Install as a Claude Code plugin or download a self-contained binary from GitHub Releases. Configure an MCP client (Claude Code, VSCode, or Claude Desktop) to connect via stdio or HTTP. Then describe a task in plain English to an AI agent, which calls WinWright tools to execute it. Tools can also be replayed deterministically with winwright run <script.json> or repaired with winwright heal.

Key features of WinWright

Record AI sessions as portable JSON scripts
Replay scripts without an AI agent or LLM calls
Self-heal broken UI selectors automatically
~59 consolidated tools with action/mode parameters
Permissions-based security, dangerous ops disabled by default
Supports both stdio and HTTP MCP transports

Use cases of WinWright

Scripted UI test automation for CI pipelines
Autonomous desktop automation via AI agents
Data extraction from legacy enterprise apps without an API
Cross-app workflows spanning desktop apps and browser
Accessibility auditing by traversing the full UIA element tree

FAQ from WinWright

What operating systems does WinWright support?

Only Windows 10 or 11 (x64 or ARM64). It is not available for Linux or macOS.

Does WinWright require a .NET runtime?

No — the binary download is self-contained and does not need a .NET runtime.

How does WinWright compare to Playwright or UiPath?

WinWright is not a browser-only testing tool (use Playwright for that) nor an enterprise RPA platform (use UiPath or Power Automate). It gives AI agents access to the Windows desktop and can handle mixed desktop+browser workflows.

What are WinWright’s automation limits?

UIA reads controls one at a time, so it is not suitable for high-throughput data pipelines. For web-only testing, Playwright is more mature. For bulk data transfer, a proper API or database connection is better.

What transports and security does WinWright support?

WinWright can connect via stdio or HTTP (configurable port, default 8765). Remote administration uses five-layer security: IP allowlist, Windows Negotiate auth, AD group authorization, rate limiting, and per-user session limits. All dangerous operations are disabled by default and can be enabled via a winwright.json config file.