README Documentation
URL Reputation and Validity Checker
A FASTMCP 2.0-based service that validates URLs and checks their reputation to help identify AI hallucinations and verify web page authenticity.
Features
- URL Validation: Verify that URLs resolve to actual web pages
- Reputation Scoring: 0-100 score based on domain age, web archive presence, and technical factors
- Link Extraction: Extract and validate links from HTML or text content
- Historical Analysis: Check domain age via WHOIS and Wayback Machine
- AI Hallucination Detection: Identify patterns common in AI-generated URLs
- Caching: Redis-based caching for improved performance
- Docker Support: Easy deployment with Docker and docker-compose
Quick Start
Using Docker (Recommended)
- Clone the repository:
git clone <repository-url>
cd reputation-and-validity-checker
- Start the services:
docker-compose up -d
The MCP server will be available at http://localhost:5000
.
Local Installation
- Install dependencies:
pip install -r requirements.txt
- Start Redis (optional, for caching):
docker run -d -p 6379:6379 redis:7-alpine
- Run the server:
python -m url_reputation_checker.server
MCP Tools
check_links_reputation
Check reputation for a list of URLs.
Parameters:
urls
: List of URLs to validate
Returns:
- List of validation results with reputation scores
Example:
result = await check_links_reputation([
"https://example.com",
"https://github.com"
])
extract_and_check_links
Extract links from content and check their reputation.
Parameters:
content
: HTML or text contentcontent_type
: "html", "text", or "auto" (default: "auto")
Returns:
- Extracted links with validation results and summary
Example:
result = await extract_and_check_links(
"<a href='https://example.com'>Link</a>",
content_type="html"
)
validate_url
Validate a single URL.
Parameters:
url
: URL to validate
Returns:
- Detailed validation information
get_domain_history
Get historical information about a domain.
Parameters:
domain
: Domain name or URL
Returns:
- Domain creation date, WHOIS info, and Wayback Machine data
Reputation Scoring
The reputation score (0-100) is calculated based on:
-
Domain Age (0-30 points):
- 5+ years: 30 points
- 2-5 years: 20 points
- 1-2 years: 15 points
- 6-12 months: 10 points
- 3-6 months: 5 points
- <3 months: 2 points
-
Web Archive Presence (0-20 points):
- 100+ snapshots: 20 points
- 50-99 snapshots: 15 points
- 20-49 snapshots: 10 points
- 5-19 snapshots: 5 points
- 1-4 snapshots: 2 points
-
Technical Factors (0-25 points):
- Valid SSL: 10 points
- Fast response (<1s): 10 points
- HTTP 200 status: 5 points
-
Consistency (0-25 points):
- No warnings: 25 points
- 1 warning: 15 points
- 2 warnings: 10 points
- 3 warnings: 5 points
AI Hallucination Detection
The service checks for patterns commonly seen in AI-generated URLs:
- Overly specific blog paths (e.g.,
/blog/2023/03/15/specific-topic
) - Version-specific API documentation URLs
- Excessive path depth (>6 levels)
- Over-hyphenated product names
- Typosquatting attempts
- Excessive subdomains
Configuration
Environment variables:
REDIS_URL
: Redis connection URL (default:redis://localhost:6379
)MCP_SERVER_HOST
: Server host (default:0.0.0.0
)MCP_SERVER_PORT
: Server port (default:5000
)
Development
- Install development dependencies:
pip install -e ".[dev]"
- Run tests:
pytest
- Format code:
black url_reputation_checker/
- Lint code:
ruff url_reputation_checker/
Docker Build
Build the Docker image:
docker build -t url-reputation-checker .
Run with docker-compose:
docker-compose up -d
View logs:
docker-compose logs -f
Example Usage
Python Client Example
import asyncio
from fastmcp import FastMCPClient
async def main():
async with FastMCPClient("http://localhost:5000") as client:
# Check a list of URLs
results = await client.call_tool(
"check_links_reputation",
urls=[
"https://github.com",
"https://example-fake-site-12345.com"
]
)
for result in results:
print(f"{result['url']}: {result['reputation_score']}/100")
# Extract and check links from HTML
html_content = """
<html>
<body>
<a href="https://google.com">Google</a>
<a href="https://fake-news-site.com">Fake News</a>
</body>
</html>
"""
extraction_result = await client.call_tool(
"extract_and_check_links",
content=html_content,
content_type="html"
)
print(f"Found {extraction_result['summary']['total_links']} links")
print(f"Average reputation: {extraction_result['summary']['average_reputation_score']}")
if __name__ == "__main__":
asyncio.run(main())
API Endpoints
The MCP server exposes tools that can be called via the MCP protocol. Additionally, it provides resources:
/url_validation_report
: Get a formatted report of all validated URLs/cache_stats
: Get cache statistics
Troubleshooting
Redis Connection Issues
If Redis is not available, the service will continue to work without caching. To disable Redis entirely, set:
export REDIS_URL=""
WHOIS Lookup Failures
Some domains may not have WHOIS information available due to privacy protection or registry limitations. The service will continue with other checks.
Wayback Machine Rate Limiting
The Wayback Machine API may rate limit requests. The service handles this gracefully and will skip historical data if unavailable.
License
MIT License - see LICENSE file for details.