Email Verifier (SMTP, MX, Catch-All)

English | 简体中文

Bulk Email Verification API — for n8n, AI Agents & MCP

Verify email addresses in bulk: RFC 5322 syntax → DNS/MX → disposable, role and free-provider lists → a live SMTP handshake with a mandatory catch-all probe. No email is ever sent — the check opens an SMTP session, reads the reply codes and hangs up.

➜ Try it on Apify

{

"email": "john@example.com", "status": "deliverable", "sub_status": "valid_mailbox", "score": 100, "checks": { "syntax": true, "domain_exists": true, "mx_found": true, "smtp_connectable": true, "mailbox_exists": true, "catch_all": false, "disposable": false, "role_based": false, "free_provider": false, "a_record_fallback": false, "has_tag": false }, "mx_record": "aspmx.l.google.com" }

Why pipelines like it

Most verifiers are written for a human reading a dashboard. This one assumes the caller is code.

That changes what the output has to guarantee:

1. `status` is always one of four values. deliverable, undeliverable, risky, unknown. Never null, never empty, never a fifth value. Enumerate them once and stay correct.

2. All eleven `checks` keys are present on every record. A check that never ran is false, and the reason lives in sub_status. Nothing disappears.

3. Catch-all domains and Gmail-type providers are reported honestly. They come back risky with a reason, not a fabricated "valid".

Those three together mean you write this:

const mailable = results.filter(r => r.status === 'deliverable');

const neverSend = results.filter(r => r.status === 'undeliverable' || r.checks.disposable); const retryLater = results.filter(r => r.status === 'unknown');

and not this:

const mailable = results.filter(r => r?.status === 'valid' && (r.checks?.catch_all ?? false) === false);

The second line is what every nullable-by-default API forces on you. It matters more for an agent than for a person: an agent cannot write `?? false` — it acts on what it reads. A field that sometimes vanishes is not an inconvenience, it is a wrong answer delivered with confidence.

What to do with each result

Full table with scores and reasoning: [docs/statuses.md](docs/statuses.md)

The distinction that matters: undeliverable means *we found out it is dead*; unknown means *we could not find out*. Tools that merge the two report better-looking numbers and quietly delete real customers from your list.

Use it in n8n

Every template calls the Actor with the core HTTP Request node — no community node to install, works on n8n Cloud and self-hosted. Authentication is an n8n Header Auth credential named Apify API; no token is stored in any file here.

Chinese versions of all four, with localized sticky notes: workflows/zh-CN/

To import: download the JSON, then in n8n use *Workflows → Import from File*. Setup steps are on a sticky note inside each workflow.

Use it from an AI agent

MCP — the Apify MCP server, scoped to this one Actor. Config for Claude Desktop and Cursor: agents/mcp/

OpenAPI — one file that Dify, Coze, GPT Actions and LangChain's OpenAPI toolkit can all import: openapi/email-verifier.openapi.yaml

Tool definitions — ready-made JSON for OpenAI function calling and the Anthropic tool API: agents/tool-definitions/

For the agent itself — llms.txt is the whole contract in one pass: endpoint, input, output, the action table, batching limits.

Chinese-first setup guides: Dify · Coze / 扣子

Use it from code

curl -X POST "https://api.apify.com/v2/acts/sumitr_mardy~email-verifier/run-sync-get-dataset-items" \

-H "Authorization: Bearer $APIFY_TOKEN" \ -H "Content-Type: application/json" \ -d '{"emails":["john@example.com","someone@gmail.com"]}'

Python and Node examples, each under 25 lines and with no SDK required: examples/

Large lists

Addresses go in one batch per call, never one call per address — batching is what makes this cheap and fast.

A synchronous Apify run must finish inside 300 seconds. Past that, use the async pattern: start the run, wait for it, then fetch the dataset. Working example: examples/batch-large-list/

Measured throughput and the batch size to use: docs/pricing-and-batching.md

Pricing

Pay per event: one charge per address that gets a verdict — including undeliverable and unknown, because the work was done either way, and a confirmed bad address is exactly the result you were paying for. No per-run charge.

The rate is tiered by your Apify plan and is adjusted to track the market, so it is not reprinted here where it would go stale: [see current pricing](https://apify.com/sumitr_mardy/email-verifier/pricing).

You need an Apify account to call the API; there is a free tier.

Known limits

Stated plainly, because you would find them anyway.

• Free providers cannot be mailbox-verified. Gmail, Outlook, Yahoo and iCloud answer 250 for addresses that do not exist, then bounce at delivery. Any vendor claiming a per-mailbox Gmail verdict from SMTP is guessing. You get risky/free_provider_unverifiable and checks.mailbox_exists: false.
• Catch-all domains cannot be mailbox-verified. By definition. The probe detects them and says so rather than returning a confident guess.
• Greylisting is not a verdict. 421/450 come back as unknown/greylisted, which is retryable. Template T3 does the retry for you.

FAQ

Does it send an email? No. It opens an SMTP session, sends EHLO, MAIL FROM and RCPT TO, reads the reply codes and disconnects. It never issues DATA, so no message body exists anywhere in the process.

Can it verify Gmail addresses? It verifies that the syntax and domain are valid and that the address is not disposable. It cannot confirm that a specific Gmail mailbox exists — Google answers identically for real and fake addresses. Nothing that uses SMTP can do better; tools that claim otherwise are guessing.

What is a catch-all domain? A domain configured to accept mail for every address, existing or not. Before asking about your address, the verifier sends a RCPT TO for a random local part that cannot exist. If that is accepted, the domain is catch-all and every address on it is reported risky/catch_all.

How many emails per call? As many as finish inside the 300-second sync limit. Measure with your own list; see docs/pricing-and-batching.md. Beyond that, use the async pattern.

How do I use it with an Apollo / Instantly / Lemlist export? Template T3 — upload the CSV, get three cleaned lists back. It picks the email column automatically and deduplicates first, so you are not charged twice for the same address.

What happens to duplicates? They are charged. Deduplicate before you call; every template here already does.

MIT licensed. Issues and pull requests welcome — particularly corrections to the n8n templates from anyone running a version where something behaves differently.