Home › Browser Automation & Scraping › bun-win32
bun-win32
Playwright for the Windows desktop, from Bun — drive native GUIs via UI Automation + MCP.
Connect
Review any command before running it. Package names and URLs come from the server's own registry entry.
Package (npm 1.9.1)
npx -y [email protected]
Or add to your MCP client config:
{
"mcpServers": {
"bun-win32": {
"command": "npx",
"args": [
"-y",
"[email protected]"
],
"env": {
"BUN_UIA_PROFILE": "<YOUR_VALUE>",
"BUN_UIA_OS": "<YOUR_VALUE>",
"BUN_UIA_ALLOW": "<YOUR_VALUE>",
"BUN_UIA_DENY": "<YOUR_VALUE>",
"BUN_UIA_CURSOR": "<YOUR_VALUE>",
"BUN_UIA_FS_ROOT": "<YOUR_VALUE>"
}
}
}
}
BUN_UIA_PROFILE— Capability profile: 'readonly' (inspect/read only), 'safe' (read + input + window — default), or 'full' (also os + fs tools).BUN_UIA_OS— Set to '1' to allow OS-level tools (launch_app, run_program, open_path) AND filesystem tools (read_file, write_file, list_dir) regardless of profile.BUN_UIA_ALLOW— Comma-separated tool names or categories to additionally allow on top of the profile.BUN_UIA_DENY— Comma-separated tool names or categories to deny, overriding the profile and BUN_UIA_ALLOW.BUN_UIA_CURSOR— Set to 'never' to forbid the real-cursor fallback entirely (strictly cursor-free). By default clicks/drags are cursor-free but fall back to the real hardware cursor when no cursor-free path exists.BUN_UIA_FS_ROOT— Sandbox root directory that read_file/write_file/list_dir are confined to when fs tools are enabled.
Related servers
Scrapling MCP Server
Web scraping with stealth HTTP, real browsers, and Cloudflare bypass. CSS selectors supported.
UI-TARS-desktop
MCP server for web search operations
playwright-mcp
Playwright Tools for MCP
invisible_playwright_mcp
AI browser agent: browses, clicks, types, and reads real web pages from plain-English instructions.
Scrapegraph-ai
AI-powered web scraping library that creates scraping pipelines using natural language.- [ScrapeGraphAI](https://scrapegraphai.com)
Skyvern
AI-powered browser automation — navigate, click, fill forms, and extract data from any website.
Data from the Official MCP Registry