MCPInstall
crawl4ai-mcp logo

crawl4ai-mcp

by salilkadam 1 stars salilkadam/crawl4ai-mcp

Crawl4AI MCP Server enables targeted web crawling and AI content analysis.

View on GitHub
READMEmaster

Crawl4AI MCP Server

An MCP (Model-Controller-Processor) Server for intelligent web crawling and AI-powered content analysis. This server provides a simple API for crawling websites and processing the content using Claude AI models.

Who Benefits from Crawl4AI?

Crawl4AI is designed for individuals and organizations who need targeted, in-depth analysis of specific web content. Unlike general search engines or AI assistants that provide broad coverage, Crawl4AI offers deeper insights into content you specifically want to analyze.

Ideal for:

  • Researchers who need to extract structured information from specific websites or academic resources
  • Content creators looking to analyze competitor content or industry trends within specific domains
  • Data analysts who need to process web data for business intelligence purposes
  • Developers building applications that require web content analysis capabilities
  • Digital marketers analyzing industry websites, blogs, or competitor content
  • Business analysts gathering industry-specific information from multiple sources
  • Knowledge workers who need to synthesize information from specific web domains

How Users Benefit from Crawl4AI

The Crawl4AI MCP server provides significant advantages over general-purpose search and AI tools:

  • Targeted depth over breadth: Instead of broad surface-level results across the entire web, get comprehensive analysis of specific websites that matter to you
  • Customizable crawling parameters: Control exactly how deep to crawl, what content to extract, and how to process it
  • Programmatic integration: Easily incorporate web content analysis into your own applications, workflows, and data pipelines
  • Flexible AI processing: Apply different analytical approaches to the same content - summarize, extract facts, deep analysis, or generate questions
  • Privacy and control: Keep sensitive searches and analyses private by running the server locally
  • Cost efficiency: Use your own Claude API key with precise control over token usage and processing costs
  • Automation potential: Schedule regular crawls and analyses of important websites to track changes over time
  • Customized AI prompting: Tailor the AI analysis specifically to your needs with customized prompting
  • Content transformation: Turn unstructured web content into structured, actionable information

Crawl4AI bridges the gap between simple web scraping and sophisticated AI analysis, enabling more targeted and meaningful extraction of insights from the web.

Features

  • Web crawling with customizable depth and content selectors
  • Respects robots.txt directives
  • Content extraction and processing
  • AI-powered analysis of crawled content using Claude models
  • Simple REST API
  • Configurable via command line or environment variables
  • Detailed logging

Installation

  1. Clone this repository:

    git clone https://github.com/yourusername/crawl4ai-mcp.git
cd crawl4ai-mcp

  2. Install dependencies:

    npm install

  3. Create a .env file with your Anthropic API key:

    ANTHROPIC_API_KEY=your_api_key_here

Usage

Starting the Server

Start the server with default settings:

npm start

Or use command-line options:

npm start -- --port 4000 --debug

Available options:

  • --port <number>: Port to run the server on (default: 3000)
  • --debug: Enable debug logging

API Endpoints

Crawl a Website

POST /api/crawl

Request body:

{
  "url": "https://example.com",
  "depth": 2,
  "selector": "main",
  "aiProcessing": {
    "task": "summarize",
    "model": "claude-3-sonnet-20240229"
  }
}

Parameters:

  • url (required): The URL to start crawling from
  • depth (optional): How many levels deep to crawl (default: 1)
  • selector (optional): CSS selector for content extraction (default: "body")
  • aiProcessing (optional): Configuration for AI processing
    • task: Type of processing (summarize, extract, analyze, questions)
    • model: Claude model to use (default: "claude-3-sonnet-20240229")

Health Check

GET /api/healthcheck

Returns server status and version information.

AI Processing Tasks

The server supports several AI processing tasks:

  • summarize: Create a comprehensive summary of the crawled content
  • extract: Extract factual information from the content
  • analyze: Perform deep analysis of the content, arguments, and quality
  • questions: Generate important questions and answers based on the content

Configuration

You can configure the server using environment variables:

  • PORT: Server port (default: 3000)
  • ANTHROPIC_API_KEY: Your Anthropic API key for Claude
  • DEBUG: Set to "true" to enable debug logging

Example

Crawl a website and summarize its content:

curl -X POST http://localhost:3000/api/crawl \
  -H "Content-Type: application/json" \
  -d '{
    "url": "https://example.com",
    "depth": 1,
    "aiProcessing": {
      "task": "summarize"
    }
  }'

License

MIT License

Acknowledgements

This project uses the following libraries:

CI/CD Setup

This project uses GitHub Actions for CI/CD and ArgoCD for GitOps-based deployment to Kubernetes.

Prerequisites

  1. A Kubernetes cluster with ArgoCD installed
  2. A Docker registry account
  3. GitHub repository access
  4. GitHub CLI installed locally

Setup Steps

  1. Configure GitHub Secrets

    # Make the setup script executable
chmod +x scripts/setup-secrets.sh

# Run the script to set up GitHub secrets
./scripts/setup-secrets.sh

  2. Deploy ArgoCD Application

    # Apply the ArgoCD application manifest
kubectl apply -f k8s/argocd/applications/crawl4ai-mcp.yaml

  3. Verify Deployment

    # Check ArgoCD application status
argocd app get crawl4ai-mcp

# Check pod status
kubectl get pods -n crawl4ai

CI/CD Pipeline

The pipeline consists of two main stages:

  1. Build and Push

    • Triggered on push to main branch and pull requests
    • Builds Docker image using the Dockerfile
    • Pushes image to Docker registry with appropriate tags

  2. Deploy

    • Triggered only on push to main branch
    • Updates ArgoCD application with new image tag
    • Triggers ArgoCD sync to deploy to Kubernetes

Directory Structure

.
├── .github/
│   └── workflows/
│       └── ci-cd.yml           # GitHub Actions workflow
├── k8s/
│   ├── argocd/
│   │   └── applications/
│   │       └── crawl4ai-mcp.yaml  # ArgoCD application manifest
│   └── base/
│       ├── deployment.yaml     # Kubernetes deployment
│       ├── service.yaml        # Kubernetes service
│       └── kustomization.yaml  # Kustomize configuration
├── scripts/
│   └── setup-secrets.sh       # GitHub secrets setup script
└── Dockerfile                 # Container build definition

Security Considerations

  1. All sensitive credentials are stored as GitHub secrets
  2. Kubernetes secrets are managed through ArgoCD
  3. Docker images are tagged with commit SHA for traceability
  4. Automated sync policies ensure cluster state matches Git repository

Troubleshooting

  1. GitHub Actions Failures

    • Check Actions tab in GitHub repository
    • Verify all required secrets are set
    • Check Docker registry access

  2. ArgoCD Sync Issues

    • Check ArgoCD UI for sync status
    • Verify Kubernetes cluster access
    • Check application logs in ArgoCD

  3. Deployment Problems

    • Check pod logs: kubectl logs -n crawl4ai <pod-name>
    • Verify service: kubectl get svc -n crawl4ai
    • Check events: kubectl get events -n crawl4ai