All articles
    MCP

    Email Verification Inside Claude and ChatGPT: Using an MCP Server

    Connect the VeriMailX MCP server to Claude, ChatGPT or Cursor and verify email addresses, clean lists and check disposable domains without leaving the chat. Setup, tools and example prompts.

    VeriMailX Team August 11, 2026 9 min read
    Email Verification Inside Claude and ChatGPT: Using an MCP Server

    Key takeaways

    • The VeriMailX MCP server exposes five tools: verify_email, verify_email_list, check_disposable_domain, get_credit_balance and list_validation_jobs.
    • It is a remote, OAuth-protected server — add the URL, sign in to your VeriMailX account, and the assistant acts as you.
    • Verification runs real SMTP-level checks, not a regex, so results match what the dashboard and API return.
    • Disposable-domain lookups are free; single and batch verification consume credits from your own balance.
    • Best for lead lists, CRM cleanup and signup-data triage done inside an AI workflow.

    Language models cannot check whether an email address exists. There is no way to reason your way to an MX record or an SMTP response — it requires a real network call. That is exactly the gap an MCP server fills, and email verification turns out to be one of the most natural things to hand an assistant.

    What you can do once it is connected

    • Paste 30 leads from a spreadsheet and ask which ones are safe to email.
    • Ask whether a signup domain is a disposable or throwaway provider before granting a trial.
    • Have an assistant clean a CRM export and explain why each address was dropped.
    • Check your remaining credit balance and recent verification jobs without opening the dashboard.

    The assistant handles the messy part — extracting addresses from pasted text, deduplicating, summarising — while the MCP server does the part that needs real infrastructure.

    The tools the VeriMailX MCP server exposes

    • verify_email — full SMTP-level check on one address. Returns valid, invalid, risky, catch-all, disposable or role-based, plus the underlying signals. One credit.
    • verify_email_list — verifies up to 20 addresses in a single call and stops early if credits run out.
    • check_disposable_domain — free lookup against our blocklist of 8,200+ throwaway email domains.
    • get_credit_balance — current plan, credits remaining and monthly usage.
    • list_validation_jobs — your recent verification jobs and their status.

    Results are identical to the dashboard and the REST API, because all three call the same verification engine.

    Connecting it

    The server is remote and OAuth-protected, so there is nothing to install:

    • Open your client's connector or MCP settings — Claude, ChatGPT, Cursor, Windsurf and other MCP-compatible assistants all support remote servers.
    • Add the VeriMailX MCP endpoint URL.
    • Approve the sign-in prompt. If you are signed out, you will land on the VeriMailX login page and return to the consent screen automatically.
    • The tools appear in the assistant. Every call from then on runs as your account, using your credits.
    New accounts include free validations, so you can test the whole flow before spending anything.

    Example prompts that work well

    • *"Here are 24 emails from our webinar signup sheet. Verify them and give me a clean list plus a table of what was rejected and why."*
    • *"Is protonmail.com a disposable provider? What about mailinator.com and tempmail.dev?"*
    • *"Check these five sales leads and flag any catch-all domains so I know which replies to expect."*
    • *"How many verification credits do I have left this month, and what did my last three jobs process?"*
    • *"Verify this address and explain in one line whether I should trust it for a cold outreach sequence."*

    Vague prompts work too. You do not need to name the tool — that is the point of MCP.

    What the results mean

    • Valid — the mailbox accepted verification. Safe to send.
    • Invalid — the mailbox does not exist. Remove it; these are the addresses that cause hard bounces.
    • Catch-all — the domain accepts everything, so the specific mailbox cannot be confirmed. Segment separately.
    • Disposable — a throwaway provider. Usually block at signup.
    • Role-based — info@, support@, sales@. Often fine for B2B, poor for personal outreach.
    • Risky / unknown — the server did not give a clean answer. Treat with caution and re-check later.

    Ask the assistant to group results by category and it will; that grouping is normally where the time goes when cleaning a list by hand.

    When to use bulk upload instead

    The MCP batch tool is capped at 20 addresses per call on purpose. Pushing thousands of rows through a chat interface wastes context and money. For anything larger, upload the CSV in the dashboard: the bulk pipeline preserves your original columns, runs in the background, and exports a filtered file when it finishes.

    A good split: use the assistant for triage, judgment calls and small lists; use bulk upload for real list hygiene before a campaign.

    Why this beats copy-pasting into a web tool

    • No context switching — the verification happens where the data already is.
    • The assistant explains the results rather than dumping a status code.
    • Follow-up questions ("now draft outreach only to the valid B2B ones") work immediately.
    • Everything is metered on your own account, so usage stays visible and auditable.

    Further reading

    Frequently asked questions

    Ready to clean your list?

    Verify your emails with VeriMailX and send your next campaign with more confidence, fewer bounces and better results. Unlimited free single email verification — no card required.

    Keep reading

    MCP

    What Is an MCP Server? A Plain-English Guide to Model Context Protocol

    An MCP server lets AI assistants like Claude, ChatGPT and Cursor call real tools and read real data. Here is what Model Context Protocol is, how an MCP server works, and when you need one.

    Read
    MCP

    How to Build an MCP Server: A Step-by-Step Developer Guide

    A practical walkthrough for building an MCP server — choosing a transport, defining tools and schemas, handling auth and errors, testing with MCP Inspector, and shipping it to Claude, ChatGPT and Cursor.

    Read
    MCP

    Remote MCP Servers and OAuth: How Authentication Works

    Remote MCP servers let anyone connect a hosted tool to Claude, ChatGPT or Cursor with a URL. Here is how OAuth 2.1, dynamic client registration and per-user scoping keep those connections safe.

    Read