Puppeteer

A comprehensive Model Context Protocol (MCP) server that provides browser automation capabilities using Puppeteer. This server enables Large Language Models (LLMs) to interact with web pages, take screenshots, and execute JavaScript in both local and cloud environments.

- Browser Automation: Navigate, click, fill forms, and interact with web pages
- Screenshot Capture: Take full-page or element-specific screenshots
- JavaScript Execution: Run custom scripts in browser context
- Console Monitoring: Capture and access browser console logs
- Resource Management: Store and retrieve screenshots and logs
- Security Controls: Configurable dangerous argument filtering

- Local Development: Direct Puppeteer integration with Chromium
- Docker Container: Containerized deployment with ARM64 support
- Cloudflare Workers: Cloud deployment using external browser services

Option 1: NPX (Recommended for Local Development)

npx -y @modelcontextprotocol/server-puppeteer

docker run -i --rm --init -e DOCKER_CONTAINER=true puppeteer-mcp

git clone https://github.com/code-craka/puppeteer-mcp.git cd puppeteer-mcp npm install npm run build node dist/index.js

- puppeteer_navigate- Navigate to URLs with optional launch options
- puppeteer_click- Click elements using CSS selectors
- puppeteer_hover- Hover over elements
- puppeteer_fill- Fill input fields and forms
- puppeteer_select- Select dropdown options

- puppeteer_screenshot- Capture page or element screenshots
- puppeteer_evaluate- Execute JavaScript in browser context

- console://logs- Access browser console output
- screenshot://<name>- Retrieve captured screenshots

Add to yourclaude_desktop_config.json:

{ "mcpServers": { "puppeteer": { "command": "npx", "args": ["-y", "@modelcontextprotocol/server-puppeteer"] } } }

{ "mcpServers": { "puppeteer": { "command": "docker", "args": ["run", "-i", "--rm", "--init", "-e", "DOCKER_CONTAINER=true", "puppeteer-mcp"] } } }

- PUPPETEER_LAUNCH_OPTIONS- JSON-encoded browser launch options
- ALLOW_DANGEROUS- Enable dangerous browser arguments (default: false)
- DOCKER_CONTAINER- Container detection flag

{ "mcpServers": { "puppeteer": { "command": "npx", "args": ["-y", "@modelcontextprotocol/server-puppeteer"], "env": { "PUPPETEER_LAUNCH_OPTIONS": "{\"headless\": false, \"args\": [\"--no-sandbox\"]}", "ALLOW_DANGEROUS": "true" } } } }

# Clone the repository git clone https://github.com/code-craka/puppeteer-mcp.git cd puppeteer-mcp # Install dependencies npm install # Development with watch mode npm run watch # Build for production npm run build

# Build Docker image docker build -t puppeteer-mcp . # Run with environment variables docker run -i --rm --init \ -e DOCKER_CONTAINER=true \ -e PUPPETEER_LAUNCH_OPTIONS='{"headless":true}' \ puppeteer-mcp

Live URL:https://puppeteer.techsci.dev

Our Puppeteer MCP server is successfully deployed on Cloudflare Workers with Browserless.io integration.

- Status: Live and operational
- Browser Service: Browserless.io connected
- API Calls: Screenshots working ✅
- Tools Available: 6 browser automation tools
- Response Time: ~1-2 seconds average

cd cloudflare-worker npm install # Installs Wrangler v4.22.0 + secure dependencies npx wrangler login echo "YOUR_BROWSERLESS_TOKEN" | npx wrangler secret put BROWSERLESS_TOKEN npm run build # Compile TypeScript to dist/index.js npm run deploy # Deploy using latest Wrangler v4.22.0

# Test tools listing curl -X POST https://puppeteer.techsci.dev \ -H "Content-Type: application/json" \ -d '{"jsonrpc":"2.0","method":"tools/list","id":"test"}' # Test screenshot capture curl -X POST https://puppeteer.techsci.dev \ -H "Content-Type: application/json" \ -d '{"jsonrpc":"2.0","method":"tools/call","params":{"name":"browser_screenshot","arguments":{"name":"test","url":"https://example.com"}},"id":"test"}'

- Cloudflare Workers: 100,000 requests/day free
- Browserless.io: ~$0.0025 per second of browser usage
- Monthly estimate: $10-50 for moderate usage

SeeCloudflare Worker READMEfor detailed setup instructions.

{ "name": "puppeteer_screenshot", "arguments": { "name": "example-page", "width": 1280, "height": 720 } }

{ "name": "puppeteer_navigate", "arguments": { "url": "https://example.com", "launchOptions": { "headless": false } } }

{ "name": "puppeteer_evaluate", "arguments": { "script": "return document.title;" } }

┌─────────────────┐ ┌──────────────────┐ ┌─────────────────┐ │ MCP Client │ │ Puppeteer MCP │ │ Browser │ │ (Claude, etc.) │◄──►│ Server │◄──►│ (Chromium) │ └─────────────────┘ └──────────────────┘ └─────────────────┘ │ ▼ ┌──────────────────┐ │ Cloud Services │ │ (Browserless.io) │ └──────────────────┘

This project is licensed under the MIT License - see theLICENSEfile for details.

- Model Context Protocol- The protocol that makes this possible
- Puppeteer- Headless Chrome Node.js API
- Anthropic- Original MCP server implementation
- Browserless.io- Cloud browser automation service

- MCP SDK- Model Context Protocol SDK
- Claude Desktop- AI assistant with MCP support
- Puppeteer- Browser automation library

- Create anissuefor bug reports
- Join discussions inGitHub Discussions
- Check theCLAUDE.mdfor development guidance

Author:Sayem Abdullah Rihan
License: MIT
Repository:github.com/code-craka/puppeteer-mcp

A configurable MCP server for browser automation using Puppeteer.