MCP Email Verification for AI Agents
By Aria Pramesi, founder of InboxPolicy · Updated November 12, 2026
InboxPolicy ships an MCP server exposing seven agent tools: decide_send, verify_email, estimate_credits, create_verify_batch, get_batch_status, get_batch_items, and get_usage. An agent asks should I email this address and gets back one of five actions (send, send_with_caution, review, retry_later, avoid), a confidence score, normalized signals, and the cost, in a single tool call, no dashboard required; agents connect with a Claude sign-in, an API key, or keyless x402 per-call payment for decide_send and verify_email.
What tools does the InboxPolicy MCP server expose?
Seven tools, all callable directly by an agent without a REST client or dashboard:
| Tool | What it returns | Inputs | Cost |
|---|---|---|---|
decide_send | One action for one address: send, send_with_caution, review, retry_later or avoid, with a reason, a reason_code, a 0 to 100 confidence, the verification status, the mail provider, a did_you_mean typo suggestion and the risk signals. | email (required), strictness, force_refresh | 1 credit |
verify_email | The full result for one address: status (valid, invalid, unknown, temporary_unavailable), deliverable and reachable flags, the signals, confidence, a reason code, the mail provider and MX hosts, the mail server's SMTP reply code, and the same recommended action decide_send gives. | email (required), strictness, force_refresh | 1 credit |
estimate_credits | operation, count and estimated_credits. An upper bound, because cache hits, duplicates, malformed addresses and unresolved results are free. | operation (verify_email, verify_batch or send_decision), count | Free |
create_verify_batch | A batch_id straight away, with status queued. Verifies up to 50,000 addresses in the background. | emails (1 to 50,000; duplicates are verified once), strictness | Reserves 1 credit per address, then refunds cache hits, duplicates, malformed addresses and unresolved results when the batch finishes |
get_batch_status | The batch's status (queued, processing, completed or failed) and summary counts. | batch_id | Free |
get_batch_items | Each address with its verification status, recommended action and signals. | batch_id, optional status filter, limit (1 to 1,000, default 100), offset | Free |
get_usage | remaining_credits, credits_used_this_month, requests_this_month, batch_jobs_this_month and your plan. | None | Free |
strictnessisdefault,aggressiveorstrict. Default sends catch-all domains toreview. Aggressive gives themsend_with_caution. Strict works like default and also sends role addresses such as info@ toreview.- A check is free when the same address was checked in the last 72 hours (unless
force_refreshis true), when the address is malformed, or when the result is unresolved (greylisted, or unknown with no catch-all or disposable signal).
How do I connect the MCP server to my agent?
The endpoint for all three options is https://mcp.inboxpolicy.com, remote MCP over Streamable HTTP, with no local install.
1. Claude (sign in, no key in the chat)
In Claude, open Customize → Connectors → Add → Add custom connector and enter https://mcp.inboxpolicy.com. Click Connect, then paste your API key on the InboxPolicy sign-in page at mcp.inboxpolicy.com and choose Approve. Claude receives an encrypted access token that lasts 30 days, never your key in plain text. Every call re-checks the key, so a revoked key stops working at once.
2. Clients that send headers
Add InboxPolicy to your agent's mcpServers config block with your API key as a header:
{
"mcpServers": {
"inboxpolicy": {
"url": "https://mcp.inboxpolicy.com",
"headers": { "x-api-key": "ip_live_..." }
}
}
}
In Claude Code, run:
claude mcp add --transport http inboxpolicy https://mcp.inboxpolicy.com --header "x-api-key: ip_live_..."
3. No key: pay per call with x402
Keyless agents can call decide_send and verify_email, and only those two, at https://mcp.inboxpolicy.com/x402. The agent sends payment in _meta["x402/payment"] and gets the receipt in _meta["x402/payment-response"]. Prices match the REST endpoints; see x402 email verification and what is an x402 API.
Get an API key by buying a credit pack at /pricing.
What does an agent get back from a single call?
Here is what decide_send returns for an address on a catch-all domain:
{
"request_id": "vrf_...",
"action": "review",
"reason": "The domain accepts mail for any address (catch-all), so this mailbox cannot be confirmed.",
"reason_code": "catch_all",
"confidence": 40,
"status": "unknown",
"mail_provider": "microsoft_365",
"did_you_mean": null,
"signals": { "catch_all": true, "disposable": false, "free": false, "role_account": false },
"from_cache": false,
"cost": { "credits": 1, "billable": true }
}
The tool result is the same JSON the REST endpoint returns, so /docs#decide and /docs#reasons describe every field. The agent does not have to parse a status enum and guess what it means for send timing. It gets the decision directly.
How does pricing work for agent calls?
Calls made with an API key or a Claude sign-in spend prepaid credits. Keyless agents pay $0.01 per call over x402, on the REST API or at the MCP server's /x402 path for decide_send and verify_email (see how to connect).
For agents or teams that prefer paying by card up front, prepaid credit packs are available: Starter $5 for 1,000 credits ($5.00/1k), Builder $19 for 5,000 ($3.80/1k), Growth $79 for 25,000 ($3.16/1k). Volume pricing is available on request. Three result types can cost 0 credits on eligible calls: a cache hit for the same normalized address and strictness within 72 hours when force_refresh is false, malformed-email rejection before SMTP, and unresolved checks (greylisted, or unknown with no catch-all or disposable signal). For API-key calls, matching idempotency replays are limited to the record TTL; keyless x402 does not use that wrapper.
There is no free tier. The $0.01 x402 call is the trial, on purpose, since free tiers tend to attract list-cleaning abuse rather than genuine agent use.
How accurate are the decisions?
Rather than claiming an unverified headline accuracy badge, InboxPolicy publishes a 40-scenario CC BY 4.0 synthetic benchmark documenting exact policy actions and known SMTP limits (including pristine spam traps).
Unknown results map to review. Catch-all results map to review by default or send_with_caution under aggressive strictness. This matters for agents because a large share of B2B addresses sit on catch-all domains, where SMTP alone cannot confirm a mailbox exists.
When should an agent use a different tool instead?
InboxPolicy is built for a live send decision inside an agent workflow, not every verification job:
- One-shot cleaning of a huge scraped list at the lowest per-address cost at high volume: MillionVerifier's $3.90/1k entry (down to $0.45/1k at 1M volume) pricing is built for exactly that. InboxPolicy's 25,000-credit Growth pack has a lower unit price at $3.16/1k, while MillionVerifier's 10,000-credit entry purchase costs less overall at $39.
- Spam-trap and abuse-address detection: ZeroBounce maintains a dedicated spam-trap and abuse database that InboxPolicy does not.
- A dashboard-first marketing suite with CSV uploads and integrations: Kickbox or Emailable are built around that workflow rather than an agent tool call.
| Tool | Entry price | Output format | MCP server | Free tier | Best for |
|---|---|---|---|---|---|
| InboxPolicy | $0.01 per call (x402), or as low as $3.16/1k on Growth credit pack | Action + confidence + normalized signals | Yes (7 tools; OAuth sign-in, API key, or keyless x402) | None (72h cache, malformed, unresolved free) | Agent send decisions in real time |
| ZeroBounce | ~$9.90/1k subscription (10k min) / ~$19.50/1k pay-as-you-go | Status + sub-status fields | Yes (official, API key) | 100 credits/mo | Spam-trap and abuse-address detection |
| Kickbox | ~$10.00/1k | Status fields plus Sendex quality score | Not documented in API docs | One-time free credits | Dashboard-first quality scoring |
| MillionVerifier | $3.90/1k entry (down to $0.45/1k at 1M volume) | Status fields, resolves unknowns aggressively | Via third-party Zapier MCP | Yes | One-shot bulk list cleaning at scale (cheapest at high volume) |
| Emailable | ~$7.60/1k pay-as-you-go (5k minimum) / ~$6.46/1k subscription | Dashboard-based verification suite | No first-party MCP; agent CLI/skill | 250 signup credits (verified July 20, 2026) | Bulk uploads and marketing integrations |
Frequently asked questions
What is MCP email verification?
MCP (Model Context Protocol) email verification means the verification service exposes its checks as tools an AI agent can call directly, instead of a REST endpoint a developer has to wire up separately. InboxPolicy's MCP server exposes decide_send, verify_email, create_verify_batch, get_batch_status, get_batch_items, estimate_credits, and get_usage, so an agent can decide whether to send an email using one tool call inside its own workflow.
Which tools does InboxPolicy's MCP server expose?
Seven tools: decide_send, verify_email, estimate_credits, create_verify_batch, get_batch_status, get_batch_items, and get_usage. decide_send and verify_email check a single address for 1 credit (returning the send action, confidence, reason code, provider details, and risk signals), create_verify_batch queues up to 50,000 addresses in the background (reserving 1 credit per address and refunding free cases on completion), and estimate_credits, get_batch_status, get_batch_items, and get_usage are free.
How much does each agent call cost?
Calls made with an API key or a Claude sign-in spend prepaid credits: Starter $5 for 1,000 credits ($5.00/1k), Builder $19 for 5,000 ($3.80/1k), or Growth $79 for 25,000 ($3.16/1k). Keyless agents pay $0.01 per call over x402, on the REST API or at the MCP server's /x402 path for decide_send and verify_email.
Can an agent use the MCP server without an API key?
Yes, for decide_send and verify_email at https://mcp.inboxpolicy.com/x402, paying $0.01 per call with x402. The batch, credit-estimate, and usage tools need an API key or a Claude sign-in.
Does InboxPolicy have a free tier?
No. The $0.01 per-call x402 price is the trial, deliberately, because free tiers tend to attract list-cleaning abuse. Three result types can cost 0 credits on eligible calls: a cache hit for the same normalized address and strictness within 72 hours when force_refresh is false, malformed-email rejection before SMTP, and unresolved checks (greylisted, or unknown with no catch-all or disposable signal).
How accurate are InboxPolicy's send decisions?
InboxPolicy runs syntax, MX, and live SMTP checks and publishes a 40-scenario synthetic routing benchmark (including documented spam-trap and catch-all edge cases) at /benchmark rather than claiming a single headline accuracy percentage. Unknown results map to review. Catch-all results map to review by default or send_with_caution under aggressive strictness; neither is silently marked safe.
When should an agent use a different email verification tool instead of InboxPolicy?
For one-shot cleaning of a large scraped list at the lowest cost at high volume, MillionVerifier is the budget option at $3.90/1k at entry, down to $0.45/1k at 1M volume (InboxPolicy's 25,000-credit Growth pack is $3.16/1k; MillionVerifier's 10,000-credit entry purchase costs less overall at $39). For spam-trap and abuse-address detection, ZeroBounce carries a dedicated database InboxPolicy doesn't maintain. For a dashboard-first suite built around CSV uploads, Kickbox or Emailable fit better than an agent-first API.