My AI agents were signing up for services with invalid emails. Disposable domains. Role accounts. Typoed MX records. No bounce handling. Just silent failures.
Then I wired up email verification into the agent toolchain.
npx @rumblingb/email-verify-mcp
One command. No API keys. No third-party fees. Pure DNS + SMTP verification.
What it checks
- RFC 5321 syntax validation
- MX record existence
- SMTP handshake (inbox exists?)
- Disposable email detection (27+ burner domains)
- Role account flagging (noreply@, support@)
- Confidence score 0-100
The agent workflow
{
"tool": "email_verify",
"params": {
"email": "user@company.com"
}
}
Returns: { valid: true, confidence: 92, disposable: false, role: false }
Why this matters
Agents autonomously signing up for services need to know if the email is real before they store it. Otherwise your agent database fills with test@test.com and noreply@domain.com.
The numbers
- 92 npm downloads/week for email-verify-mcp
- Zero server costs (Cloudflare Workers free tier)
- Zero per-email charges
- 50 checks/day free, Pro $19/mo for 1,000
Stack
MCP server → Claude/Cursor/Copilot → agent calls verify → clean data
61 products. 26 MCP servers. Building in public at agentpay.so.
Top comments (1)
This is a good example of a small tool preventing a much bigger cleanup job.
For agents, email validation should probably be a preflight step, not a backend correction pass. Even if SMTP checks are imperfect, a confidence score plus disposable/role detection is enough to keep bad state from entering the workflow silently.