About deliverability-doctor
deliverability-doctor is a remote MCP server that checks a domain’s email authentication posture using public DNS. It gives security grades and concrete fixes without requiring authentication.
Key features
- 01Check SPF, DMARC, MX, and DKIM
- 02Grade results from A to F
- 03Report findings with suggested fixes
- 04Compare two to five domains
Use cases
- 01Audit a domain for email spoofing risks
- 02Compare email security across domains
MCP support
It exposes email-security checks and domain comparisons as MCP tools over stateless streamable HTTP. It reads public DNS and requires no authentication.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://deliverability-doctor.deliverability-doctor.workers.dev/mcpclaude mcp add --transport http integrity-os-deliverability-doctor https://deliverability-doctor.deliverability-doctor.workers.dev/mcp{"mcpServers":{"integrity-os-deliverability-doctor":{"type":"http","url":"https://deliverability-doctor.deliverability-doctor.workers.dev/mcp"}}}Open in Cursor{"servers":{"integrity-os-deliverability-doctor":{"type":"http","url":"https://deliverability-doctor.deliverability-doctor.workers.dev/mcp"}}}Open in VS Code{"mcpServers":{"integrity-os-deliverability-doctor":{"type":"http","url":"https://deliverability-doctor.deliverability-doctor.workers.dev/mcp"}}}{"mcpServers":{"integrity-os-deliverability-doctor":{"serverUrl":"https://deliverability-doctor.deliverability-doctor.workers.dev/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
2| Parameter | Type | Required | Description |
|---|---|---|---|
| domain | string | Yes | Domain, URL, or email address, e.g. example.com |
Details
From the project READMEDeliverability Doctor
Remote MCP server: "Can my domain be email-spoofed?" answered from public DNS, graded, with concrete fixes. Connectable to Claude (custom connector) and ChatGPT (Apps SDK / MCP connector) — both speak MCP, one server serves both.
Tools
check_email_security(domain)— SPF (incl. RFC 7208 multiple-record permerror andredirect=handling), DMARC (policy, rua, pct), MX, DKIM at 7 common selectors. Graded A–F report, fixes per finding, upsell line only when problems exist. Accepts bare domain, URL, or email address.compare_email_security(domains[2..5])— comparison table with grades.
Design decisions
- Authless: reads public DNS only, stores nothing, takes nothing but a domain. Lowest possible connect friction; nothing sensitive to protect.
- DNS-over-HTTPS (Cloudflare → Google fallback): identical code runs on local
Node and on Cloudflare Workers — no
dnsmodule dependency. - Stateless streamable HTTP: fresh transport per request, no sessions; safe to scale, trivial for clients.
- NXDOMAIN is refused, not graded — "domain doesn't exist" ≠ "spoofable".
- Verified against known postures before first run: gmail.com (
redirect=must not be flagged), n8n.io (must grade A onp=reject), missing-everything domain (must grade F), NXDOMAIN (must refuse).
Run
npm install
node server.js # :8787/mcp
Quick public demo: cloudflared tunnel --url http://localhost:8787 → use
https://<random>.trycloudflare.com/mcp as the connector URL.
Status 2026-08-23
- Local: end-to-end MCP verified (initialize / tools/list / tools/call).
- Public: live via quick tunnel, initialize verified from the public URL.
- Durable hosting: pending — Cloudflare Workers deploy needs Kacper's account (~5 min). Tunnel URL dies with the process/PC.
- Directory submissions (Anthropic connector directory, ChatGPT apps): need the durable URL first, then Kacper's developer accounts.
Monetization path
Free tool = distribution. Report links the $99 written audit (niekonieczny.gumroad.com/l/vscjt) only when problems are found. NOTE: Gumroad payouts currently frozen on Stripe KYC — unfreeze before promoting.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.