About Verifly
Verifly is an MCP server for AI agents and other MCP clients to verify email addresses and manage email-list workflows. Its tools cover single and bulk checks, account onboarding, and credit usage.
Key features
- 01Real-time single-address verification
- 02Synchronous batch verification
- 03Async bulk jobs and results
- 04List deduplication and cleaning
- 05Account registration and credit tracking
Use cases
- 01Check a lead before sending email
- 02Clean and deduplicate an import list
- 03Run and track a large verification job
MCP support
Exposes 15 email-verification and account-management tools to MCP clients. It is available as a local stdio server or hosted Streamable HTTP endpoint.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://verifly.email/mcpclaude mcp add --transport http email-verifly-verifly https://verifly.email/mcp{"mcpServers":{"email-verifly-verifly":{"type":"http","url":"https://verifly.email/mcp"}}}Open in Cursor{"servers":{"email-verifly-verifly":{"type":"http","url":"https://verifly.email/mcp"}}}Open in VS Code{"mcpServers":{"email-verifly-verifly":{"type":"http","url":"https://verifly.email/mcp"}}}{"mcpServers":{"email-verifly-verifly":{"serverUrl":"https://verifly.email/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
https://verifly.email/mcpAuth requiredTools
17This tool takes no parameters.
Details
From the project README
Free email tools & API
- Catch-all email verifier — check if a domain is accept-all
- Disposable email checker
- Bulk email verification API — clean CSV lists
- Email verification API for developers — pay-as-you-go, 100 free credits, no monthly fee
Verifly MCP Server
Agent-native email verification over the Model Context Protocol. This server gives any MCP client — Cline, Claude Desktop, Claude Code, Cursor, Windsurf — a clean set of tools to verify email addresses, clean lists, run async bulk jobs, and manage a Verifly account, all without writing a single raw HTTP call.
It is built for autonomous workflows: an agent can self-register for an API key, verify a lead before sending, deduplicate and scrub an import list, kick off a bulk job and poll it to completion, and check its own credits and usage — end to end.
Tools
15 tools cover the full Verifly workflow:
| Tool | What it does |
|---|---|
verify_email | Verify a single address in real time (verdict, reason, flags, send/reject recommendation, credits) |
verify_batch | Verify a list synchronously (per-address verdicts; best for up to a few hundred) |
clean_email_list | Dedupe + drop invalid syntax, disposable, and role addresses from a list |
extract_emails | Pull every email address out of free-form text (notes, signatures, pasted docs) |
check_domain_health | MX / SPF / DMARC records + an overall health score for a domain |
get_credits | Remaining verification credits + recent usage (free, no credits) |
submit_bulk | Submit an async bulk verification job for large lists (returns a job_id; optional webhook_url) |
get_job_status | Status + progress of a bulk job |
get_job_results | Per-address results of a completed bulk job |
list_jobs | List the account's bulk jobs (filter by status, paginate) |
get_usage | Detailed usage statistics for a period (day / week / month) |
get_account | Account profile: email, company, credits, plan, key count (free) |
get_packages | List the credit packages available to purchase |
buy_credits | Start a credit-package purchase and return a checkout link |
register_account | Self-onboard a brand-new account; returns an api_key shown once |
Authentication
Get an API key from https://verifly.email (or have an agent self-register with
register_account). The server reads the key from the VERIFLY_API_KEY
environment variable. Every Verifly request is authenticated with
Authorization: Bearer <key>.
The hosted HTTP transport also accepts a per-request Authorization: Bearer <key>
header (which overrides the env var), so each caller can supply their own key.
Install (local, stdio) — recommended for Cline
npx -y verifly-mcp-server
or install it globally:
npm install -g verifly-mcp-server
verifly-mcp-server
Client config
Add this to your MCP client config (Cline: cline_mcp_settings.json; Claude
Desktop: claude_desktop_config.json):
{
"mcpServers": {
"verifly": {
"command": "npx",
"args": ["-y", "verifly-mcp-server"],
"env": {
"VERIFLY_API_KEY": "vf_your_api_key"
}
}
}
}
That's the whole setup — no build step, no clone. See
llms-install.md for an agent-followable install guide.
Hosted Streamable-HTTP transport
If you'd rather not run anything locally, Verifly hosts the same server as a remote Streamable-HTTP endpoint:
https://verifly.email/mcp
Point any Streamable-HTTP-capable MCP client at that URL and pass your key as an
Authorization: Bearer <key> header. Example raw call:
curl -s -X POST https://verifly.email/mcp \
-H "Content-Type: application/json" \
-H "Accept: application/json, text/event-stream" \
-H "Authorization: Bearer vf_your_api_key" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call",
"params":{"name":"verify_email","arguments":{"email":"[email protected]"}}}'
Self-hosting the HTTP transport
PORT=8787 VERIFLY_API_KEY=vf_your_api_key node src/http.js
# or: PORT=8787 VERIFLY_API_KEY=vf_... npm run start:http
| Var | Default | Purpose |
|---|---|---|
PORT | 8787 | Listen port |
HOST | 0.0.0.0 | Bind address |
MCP_PATH | /mcp | Path the MCP endpoint is served on |
VERIFLY_API_KEY | — | Fallback key when a request sends no Authorization header |
VERIFLY_BASE_URL | https://verifly.email | API base override |
- MCP endpoint:
POST {MCP_PATH}(stateless Streamable HTTP). - Health check:
GET /healthz→{"ok":true}.
Test
VERIFLY_API_KEY=vf_your_api_key npm run test:e2e
Drives the stdio server through the MCP protocol and performs a real
verify_email + get_credits call against the live API.
License
MIT
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.