Let your AI assistantcheck a domain.
A free Model Context Protocol server. Connect it to ChatGPT, Claude, Cursor or any MCP client, and your assistant can check any domain's email security, website security and DNS itself, with the same results as the checks on this site.
In ChatGPT or Claude
Open Settings, then Connectors, add a custom connector and paste:
https://fyzno.com/mcpNo account, no key. Custom connectors are a paid-plan feature in both apps.
Locally: Claude Code, Claude Desktop, Cursor, Windsurf, Zed
Add this to the tool’s MCP configuration:
{
"mcpServers": {
"fyzno": { "command": "npx", "args": ["-y", "@fyzno/mcp"] }
}
}Published on npm as @fyzno/mcp, source on GitHub. Same tools, identical results.
Seventeen tools, all read-only
- mail_health_check
- The full mail report for a domain, scored out of 100.
- web_security_check
- HTTPS and redirects, certificate, TLS, HSTS, Content-Security-Policy, framing and cookie flags, scored out of 100.
- dns_hygiene_check
- Name server redundancy, CAA, SOA timers and wildcards, scored out of 100.
- check_spf · check_dkim · check_dmarc
- SPF record and lookup count, DKIM keys, DMARC policy and alignment.
- check_mta_sts · check_tls_rpt · check_smtp_tls
- MTA-STS policy over HTTPS, TLS reporting, and a live STARTTLS handshake.
- check_routing · check_certificates · check_reputation
- MX routing, the certificates on the mail servers, and the Spamhaus listing.
- check_bimi · check_ipv6
- Brand logo record and IPv6 reachability.
- site_status · site_data · pages
- Fyzno's own status and public facts, for agents that ask about us.
Then just ask
- “Check the mail setup for example.com.”
- “Why would mail from example.com land in spam?”
- “Run a website security check on example.com.”
- “Does example.com have a CAA record?”
What it can do to you
Nothing. Every tool is a read, and the only input is a domain name. Nothing is stored. The web check fetches a home page the way a browser would and sends no payload; the DNS check reads published records only. Checks are rate-limited per address, so one agent cannot use it to hammer someone else’s servers.