MCP Domain Lookup Server

A Model Context Protocol server that provides DNS and domain intelligence tools. Designed to run as a remote MCP server over Streamable HTTP transport and publish on Smithery.ai.

Tools

Tool	Description
dns_lookup	Look up DNS records (A, AAAA, MX, TXT, NS, CNAME) for any domain
whois_lookup	Get WHOIS registration data — registrar, dates, expiry countdown
domain_available	Check if a domain is available; suggests alternatives if taken
email_config_check	Audit email security (MX, SPF, DKIM, DMARC) with A–F grading
ssl_check	Inspect SSL/TLS certificate — issuer, expiry, trust status

Quick Start

npm install
npm start

The server starts on port 3000 by default (override with PORT env var).

Endpoints

Method	Path	Purpose
POST	/mcp	MCP Streamable HTTP (JSON-RPC)
GET	/mcp	SSE stream for existing sessions
DELETE	/mcp	Session cleanup
GET	/.well-known/mcp/server-card.json	Server metadata card
GET	/health	Health check
GET	/	Server info

Deploy

Works on any Node.js host (Railway, Render, Fly.io, etc.):

1. Set the PORT environment variable if needed (most platforms inject it automatically).
2. Start command: node server.js

No API keys or external services required — all lookups use built-in DNS resolution, raw WHOIS queries, and direct TLS connections.

Observability / Usage Logs

The server emits structured JSON logs for every request so you can see what is waking or using the service in Render logs.

Logged events include:

• http_request — every HTTP request with method, path, status, duration, client IP, selected headers, and query parameters
• mcp_post — MCP JSON-RPC calls, including initialize requests and tools/call tool names/arguments
• mcp_session_initialized, mcp_stream_open, mcp_session_deleted, mcp_session_closed — MCP session lifecycle events
• mcp_discovery_requested and mcp_config_requested — discovery probes for .well-known metadata
• rest_tool_requested, rest_tool_completed, rest_tool_failed — REST API tool usage
• health_check, root_requested, page_requested — health checks and page views

Health checks are skipped by default to keep Render logs quiet. Set LOG_HEALTH_CHECKS=true if you want to include /health requests in the logs while debugging deploy or uptime behavior.

Each request gets an x-request-id response header and a matching requestId in logs, making it easier to connect lifecycle/tool events to the final HTTP request log.

The logs include a best-effort sourceGuess / sourceEvidence based on user-agent, referer, origin, host, and path. This can identify likely traffic from common MCP directories/clients such as Smithery, MCP.so, MCP Store, MCP Market, PulseMCP, Glama, LobeHub, Cursor, Claude, VS Code, Windsurf, uptime checks, or generic bots. If a directory or client does not send an identifying user-agent/referer, attribution may remain Unknown, MCP client, or MCP discovery probe.

Sensitive values are intentionally limited: authorization headers are not logged, MCP session IDs are redacted, and long values are truncated.

License

MIT