Headless browser automation for Cursor AI agents: scrape, navigate, fill forms, take screenshots, and run multi-step research from Cursor
概览
Headless browser automation for AI agents — scrape, crawl, search, screenshot, and run multi-step research from Cursor. This repository is the Cursor plugin for the hosted at https://mcp.browserless.io/mcp. The server speaks the Model Context Protocol over streamable HTTP, authenticated with a Bearer token. 1. Open . 2. Search for . 3. Click , then . 4. Sign in with your Browserless account and approve access. Authentication uses OAuth — there is no token to copy, paste, or rotate by hand. Cursor stores the credentials itself; nothing is written into this repository or into mcp.json. No account yet? Sign up at browserless.io/signup/email. If you are on a Cursor build older than 3.13 or prefer to manage MCP config yourself, add this to ~/.cursor/mcp.json (global) or /.cursor/mcp.json (per-project), using an API token from account.browserless.io: Restart Cursor, then check — browserless should appear with all 14 tools enumerated. See docs/install-cursor.
README
以下为原文 README 节选，完整内容请查看原文。
Browserless MCP for Cursor
Headless browser automation for AI agents — scrape, crawl, search, screenshot, and run multi-step research from Cursor.
This repository is the Cursor plugin for the Browserless MCP server hosted at https://mcp.browserless.io/mcp. The server speaks the Model Context Protocol over streamable HTTP, authenticated with a Bearer token.
Install
From the Cursor Marketplace (recommended)
- Open Cursor Settings → Plugins.
- Search for Browserless.
- Click Install, then Authenticate.
- Sign in with your Browserless account and approve access.
Or run /add-plugin browserless in chat.
Authentication uses OAuth — there is no token to copy, paste, or rotate by hand. Cursor stores the credentials itself; nothing is written into this repository or into mcp.json.
No account yet? Sign up at browserless.io/signup/email.
Manual install (fallback)
If you are on a Cursor build older than 3.13 or prefer to manage MCP config yourself, add this to ~/.cursor/mcp.json (global) or /.cursor/mcp.json (per-project), using an API token from account.browserless.io:
{
"mcpServers": {
"browserless": {
"type": "http",
"url": "https://mcp.browserless.io/mcp",
"headers": {
"Authorization": "Bearer YOUR_BROWSERLESS_TOKEN"
}
}
}
}
Restart Cursor, then check Settings → MCP — browserless should appear with all 14 tools enumerated.
See docs/install-cursor.md for deeplink installs, project-scoped config, and troubleshooting, and docs/auth.md for token management.
Tools
The hosted server exposes 14 tools and 2 read-only resources (browserless://status and browserless://api-docs). Full parameters in docs/tools.md.
Scrape & extract
| Tool | What it does |
|---|---|
browserless_smartscraper |
Fetch a single URL as markdown, html, links, screenshot, or pdf. Auto-handles JS and anti-bot. |
browserless_crawl |
Recursively crawl and scrape every discovered page. Path filters, sitemap modes, depth control. |
browserless_map |
Discover all URLs on a site via sitemap + link extraction, ranked by relevance. |
browserless_export |
Return a page’s native content (HTML/PDF/image), or bundle page + assets as a ZIP. |
browserless_function |
Run arbitrary Puppeteer JS in the cloud. Returns { data, type } with real MIME support. |
Search
| Tool | What it does |
|---|---|
browserless_search |
Web / news / image search, geo-targetable and time-filterable, with optional per-result scraping. |
Agent
| Tool | What it does |
|---|
安装
This server does not publish a one-line install command.
Open the repository installation guide 新手？先看 mcpskill.org 教程 →配置
{
"mcpServers": {
"browserless": {
"type": "http",
"url": "https://mcp.browserless.io/mcp",
"headers": {
"Authorization": "Bearer YOUR_BROWSERLESS_TOKEN"
}
}
}
}