This server gates every URL your agent encounters before it fetches content or follows a link. One tool, check_url, returns a SAFE/SUSPICIOUS/DANGEROUS verdict powered by Google Web Risk, URLhaus, PhishTank, and Claude's reasoning engine. You get a 0-100 trust score, threat categories like phishing or typosquatting, domain age from RDAP, SSL status, and a plain-English explanation of why the AI flagged it. Use it before scraping pages, processing email links, or passing URLs to browser tools. Free tier gives you 10 calls per month with no API key. Paid bundles start at $20 for 500 checks. Available as a hosted endpoint or local stdio install.
claude mcp add --transport stdio ojaskord-url-safety-validator-mcp uvx url-safety-validator-mcp