Skip to content
Start with 100 free verification credits
Qualisend
MCP · Model Context Protocol

Email verification, built for your agents.

Connect Qualisend to Claude, ChatGPT, Grok, or any MCP-compatible client and your AI agent can verify addresses and clean lists in plain language - with a real mailbox check behind every verdict. No code, no CSV exports, no copy-pasting keys.

A hosted email verification MCP server, listed in the official MCP Registry as com.qualisend/email-verification.
AI agent · Qualisend MCP
Verify sales@acme.com. Is it safe to email?
verify_email deliverable
sales@acme.com is deliverable (98/100). Real mailbox, no catch-all, not disposable - safe to send.
MCP endpoint
https://app.qualisend.com/api/mcp
Works with any MCP client

Connected in under a minute

The MCP server is a single endpoint. Add it once in your AI client and every chat can verify email from then on.

01

Add the endpoint

In your client's connector settings, paste the Qualisend MCP endpoint. That's the only URL you need.

02

Sign in or paste a key

Approve access with one-click sign-in (OAuth 2.1) or supply a Qualisend API key. The connection is scoped to the workspace you approve.

03

Just ask

“Verify this address”, “are these 50 leads deliverable?”, “how many credits are left?” The agent calls the right tool and hands back a clear verdict.

Why verify through the agent

Plain language, no code

Your agent verifies in the flow of the conversation. No SDK, no CSV export, no context-switching to a dashboard.

A real mailbox check

Every address comes back deliverable, risky, undeliverable, or unknown, with a 0-100 score. We actually ask the mail server, rather than guessing from how the address looks.

One address or a thousand

Verify a single email inline, a batch of 50 in one call, or kick off an async job for a list of up to 1,000.

One-click, scoped connect

Sign in once and approve access - there are no keys to copy or store. It stays limited to the workspace you picked, and you can cut it off whenever you like.

Credits you control

The agent uses the same credits and rate limits as the API. Ask “how many credits are left?” before a big job - reads are free.

Works everywhere MCP does

One endpoint for Claude, ChatGPT, Grok, Cursor, and any MCP-compatible client - the same verdict, wherever your agent runs.

What you can ask

There are no commands to learn. Ask in your own words and the agent picks the right tool.

“Verify sales@acme.com before I email them.”
“Are these 20 addresses deliverable? Flag anything risky.”
“Clean this list of 500 signups and tell me which to drop.”
“Is this a disposable or role address?”
“How many verification credits do I have left?”
“Did that verification job finish? Summarize the results.”

Connect it

Two ways in: sign in and approve, or send an API key. Chat assistants like Claude, ChatGPT and Grok use sign-in; developer clients take either. Either way the connection is limited to one workspace and you can revoke it at any time.

Sign in (OAuth 2.1)

Recommended

Add the endpoint as a custom connector; your client opens a “Connect Qualisend” page where you log in and approve access. No keys to copy, and you can revoke it from your client any time.

  1. 1Open your client's connector settings. In Claude that is Customize → Connectors → Add custom connector; every client has its own path, and there is a guide for each below.
  2. 2Paste the MCP endpoint above and click Connect.
  3. 3Approve access to your workspace. Done.

Or use an API key

Prefer a static credential? Developer clients - Claude Code, Cursor, VS Code, Zed, Goose and the CLIs - can send one. Create a key under Verification API → API keys in your dashboard and pass it as the bearer token (Authorization: Bearer qs_live_…). The chat assistants use sign-in instead.

MCP endpoint
https://app.qualisend.com/api/mcp

Your agent gets exactly what you approve

Connecting an AI assistant shouldn't mean handing it the keys. The same limits that protect your account through our API apply here too.

One workspace only

Your agent can only see the workspace you connected it to. Not another customer's data, not your billing or admin settings, nothing else.

The answer, nothing else

Your agent gets the verdict for an address and its score. It never sees how we check, or anything about how the service runs behind the scenes.

Nothing to hijack

The connection uses a key your agent sends with each request, not a browser login. So there's no session for a malicious page to borrow.

Off in one click

Disconnect it in your AI client, or delete the key in your dashboard. Sign-in access also expires on its own after an hour.

Full detail on how we handle your data is in the Privacy Policy.

MCP FAQ

It's a Model Context Protocol server that lets an AI agent verify email addresses and clean lists through Qualisend, in plain language. You add one endpoint to your MCP client and the agent gets a small set of verification tools - the same engine behind our REST API.

Any MCP-compatible client: Claude (on every plan, including Free), ChatGPT (through developer mode), Grok (custom connectors), Grok Bot, Cursor, VS Code, Claude Code, the Codex and Gemini CLIs, Zed, Goose, Windsurf and others. It uses standard Streamable HTTP transport, so it works the same way everywhere - only the menu path differs, and there is a setup guide for each.

No. One-click sign-in over OAuth 2.1 is the recommended path - and in the chat assistants (Claude, ChatGPT and Grok) it is effectively the only one, because their connector dialogs are built around sign-in. Developer clients like Claude Code, Cursor, VS Code and the CLIs also accept a Qualisend API key as the bearer token.

Only the verification tools, scoped to the single workspace you connect. It can verify addresses, run list jobs, and read your credit balance and job results - nothing else. It cannot reach another customer's data, admin functions, or any internal system.

Verifying an address costs one credit (the same as the API) whether the agent checks one or a thousand. Reads like checking your credit balance or a job's results are free. You can start on the free plan with 100 credits that never expire.

Yes. Your agent only ever sees the verdict for an address, never anything about how we check it or how the service runs. The connection uses a key sent with each request rather than a browser login, so there is no session for a malicious page to borrow, and you can cut access off at any time from your AI client or your dashboard.

Give your agent a real deliverability check

Start free with 100 credits that never expire, connect the MCP endpoint to your AI client, and let it verify email the moment it needs to.