DeliverIQ ยท MCP server

Email deliverability tools for AI agents.

12 MCP tools. Email verification, sender reputation, DNSBL, SPF/DKIM/DMARC analysis, inbox-placement scoring. Drop into Claude Desktop in 60 seconds.

Get a free API key View on GitHub
MIT licensed Free tier included No credit card

Already have a Min8T account? Sign in to claim a key.

Add to claude_desktop_config.json

// ~/Library/Application Support/Claude/claude_desktop_config.json
{
  "mcpServers": {
    "deliveriq": {
      "command": "npx",
      "args": ["-y", "@deliveriq/mcp"],
      "env": { "DELIVERIQ_API_KEY": "lc_your_key" }
    }
  }
}
5-stage pipeline

How DeliverIQ verifies every email

21 checks across 5 stages, run in parallel where possible. Sub-3-second per email.

1
Email Format Validation
Steps 1–2 · 2 checks
  • 1.1
    Syntax validation

    RFC 5322 compliance check. Validates structure, character set, and the @-placement before any network calls.

  • 1.2
    Typo detection

    Sift3 fuzzy match against a database of 1,200+ common misspellings of popular providers (e.g. gmial.comgmail.com).

2
Domain & provider checks
Steps 3–7 · 5 checks
  • 2.1
    Disposable domain detection

    Continuously maintained list of 164,000+ throwaway-email domains.

  • 2.2
    Role-based detection

    Flags 130+ role prefixes (admin@, info@, support@, noreply@…) with lower engagement and higher bounce rates.

  • 2.3
    Free-provider check

    Identifies Gmail, Yahoo, Outlook, and 200+ other consumer email domains worldwide. Useful when corporate addresses are required.

  • 2.4
    Alias normalization

    Resolves +tag and dot-aliases to canonical form (user+news@gmail.comuser@gmail.com) so duplicates collapse.

  • 2.5
    Email-pattern analysis

    Scores the local-part for entropy, keyboard walks, and bot-style sequences. Distinguishes human-typed from auto-generated.

3
Mailbox verification
Steps 8–10 · 3 checks
  • 3.1
    MX record resolution

    DNS MX lookup with A-record fallback. Highest-priority server is used for SMTP. Capped at top 2 to prevent timeout stacking.

  • 3.2
    ISP identification

    MX-pattern match against known ISPs. Major providers that block SMTP probes are skipped intelligently for accuracy and speed.

  • 3.3
    SMTP handshake verification

    Real conversation: TCP 25 connect → EHLO → MAIL FROM → RCPT TO. Second RCPT with a random address detects catch-alls. 5-second timeout. No email is sent.

4
Reputation & intelligence
Steps 11–17 · 7 checks
  • 4.1
    Gravatar lookup

    A Gravatar profile is a positive trust signal: someone set up a public identity with this address.

  • 4.2
    DNSBL blacklist check

    47 categorized zones (Spamhaus, SpamCop, SURBL, URIBL, SORBS, Barracuda, UCEProtect, DroneBL…) plus domain-based lists.

  • 4.3
    Domain age via RDAP

    Registration date, expiration, registrar, DNSSEC, nameservers. Domains under 30 days are high-risk; under a year, elevated.

  • 4.4
    HIBP breach check

    Have I Been Pwned lookup. Compromised addresses are more likely to be abandoned or spam-targeted.

  • 4.5
    DKIM record check

    Probes 15 common DKIM selectors in parallel. Returns key type (RSA / Ed25519) and estimated key size when found.

  • 4.6
    Infrastructure analysis

    SPF + DKIM + DMARC policy + MTA-STS + BIMI + TLS-RPT. Composite score out of 100. A full stack scores highest.

  • 4.7
    MX server reputation

    Reverse-DNS (PTR) and IP-DNSBL on the MX itself. Servers without PTR or with blacklisted IPs indicate lower-quality infrastructure.

5
Scoring & classification
Steps 18–21 · 4 checks
  • 5.1
    Spam-trap heuristic scoring

    12 weighted signals (age, role, no Gravatar, SMTP, disposable, DNSBL, catch-all, entropy, MX rep, IP blacklist, pattern type). Classifies trap as Pristine / Recycled / Typo with confidence.

  • 5.2
    Domain trust score

    0–100 composite: age (25) + infrastructure (25) + reputation (25) + trust signals like DNSSEC and registrar lock (25). Maps to Trusted / Positive / Neutral / Suspicious / Malicious.

  • 5.3
    Deliverability score

    Single 0–100 number combining SMTP, reputation, infrastructure, spam-trap probability, and address characteristics. The model gets one actionable verdict.

  • 5.4
    Reachability classification

    Score maps to Safe (80–100) / Risky (40–79) / Invalid (1–39) / Unknown (0). Unknown captures timeouts, greylisting, and catch-all servers.

Deliverability score scale

Every email gets a single 0–100 score, mapped to one of four categories.

80–100Safe
40–79Risky
1–39Invalid
0?
Safe · high-confidence inbox delivery
Risky · may bounce or land in spam
Invalid · almost certain bounce
Unknown · could not be determined
12 tools

Ready for the model.

Structured input, structured output. The model decides what to call when.

deliveriq_verify_email

Single-email verification. Syntax, MX, SMTP, DNSBL, SPF/DKIM/DMARC, score.

1 credit
deliveriq_batch_verify

Submit a batch for async verification. Downloadable CSV results.

1 / email
deliveriq_batch_status

Poll batch progress and per-category breakdown.

free
deliveriq_batch_download

Download finished batch as CSV. Filterable by category.

free
deliveriq_list_jobs

Paginated list of recent verification jobs.

free
deliveriq_find_email

Find a business email by name and domain. Pattern-aware.

2 credits
deliveriq_blacklist_check

Scan a domain across 50 DNSBL zones. Per-zone result.

1 credit
deliveriq_infrastructure_check

SPF, DKIM, DMARC, MTA-STS, BIMI analysis with composite score.

1 credit
deliveriq_spam_trap_analysis

Evaluate spam-trap risk for an email. 13 weighted signals.

1 credit
deliveriq_domain_intel

Composite trust report. MX, DNSBL, infra, age, ISP.

1 credit
deliveriq_org_intel

Cached organization intelligence. Email patterns, contact count.

free
deliveriq_check_credits

Plan, balance, monthly usage, daily breakdown.

free

Why MCP?

Deliverability is the kind of problem where the right answer needs 6 to 8 API calls in sequence. Verify, then DMARC, then blocklist, then sender reputation. Letting the model orchestrate as tools is roughly 10× more useful than another dashboard. It can plan, branch on results, and explain the verdict in plain English.

Free tier. No credit card.

Sign up, generate a key, paste into Claude Desktop. Five minutes from cold open to first verification.

Get my API key
Production endpoint: api.min8t.com/deliverability/v1