# Connect Claude, Cursor or VS Code to Vaartix

> Add the hosted Vaartix MCP server to Claude, Claude Code, Cursor or VS Code, sign in with your Vaartix login, and let the assistant send email and check delivery.

Source: https://vaartix.com/docs/guides/mcp-hosted/

The hosted Vaartix MCP server lets AI assistants send email and look up delivery status, emails and domains in your Vaartix account. You sign in with your normal Vaartix login and choose what the app may do. There’s no API key to copy and nothing to install.

Server URL

```text
https://mcp.vaartix.com
```

It runs in India, on the same servers as the Vaartix API (AWS Mumbai). Your emails, sign-ins and tokens never leave them. Prefer to run it yourself with an API key? See [the local server](https://vaartix.com/docs/guides/mcp/#local-npx).

## What it can do

| Tool | What it does | Permission |
| --- | --- | --- |
| `send_email` | Send one email: HTML and/or text, cc, bcc, reply-to, tags, scheduled sending, idempotency key | Send email |
| `send_batch` | Send up to 20 emails in one call | Send email |
| `cancel_email` | Cancel a scheduled email | Cancel scheduled emails |
| `get_email` | One email with its status and event timeline; its HTML and text only with **Read email content** | Read email details |
| `list_emails` | Recent emails, filtered by status, search text or number of days | Read email details |
| `list_domains` | Your sending domains, with their verification status and DNS records | Read email details |
| `get_domain` | One domain: whether it’s verified, and the DNS records it needs | Read email details |

`send_email` and `send_batch` deliver real email. Most AI apps ask you to approve each tool call: check the recipients and the content before you approve. To try it out, send to a [test address](https://vaartix.com/docs/sending/test-mode/) such as `delivered@test.vaartix.com`.

## Connect your app

**Claude**

In Claude (claude.ai, Claude Desktop or the mobile apps):

1.  Open **Customize** → **Connectors**.
2.  Select **\+ Add**, then **Add custom connector**.
3.  Name it `Vaartix` and enter the URL `https://mcp.vaartix.com`. Select **Continue**.
4.  Keep the authentication settings Claude detects: sign in, with **Use Claude’s published identity**. Select **Continue** and sign in to Vaartix.

Turn the connector on for a conversation from the **+** button → **Connectors**.

On Team and Enterprise plans, an owner adds the connector first, under **Organization settings** → **Connectors**; then each person connects and signs in with their own Vaartix login.

## Before you connect

An owner or admin turns on **Allow AI agents (MCP)** in **Settings** → **Connected apps** first. It’s off for every team until then, and while it’s off no app can connect or act for the team.

## Signing in

When the app connects, your browser opens Vaartix (log in if you aren’t already) and shows what the app is asking for:

-   **The app**, and whether Vaartix could confirm who it is. Apps that registered themselves are marked as unconfirmed: only continue if you just started connecting from that app.
-   **Where you’ll go back to**, with a warning when it’s an app on your own computer.
-   **The team** it will act on, if you’re in more than one.
-   **What it can do**: **Send email**, **Read email details**, **Read email content** and **Cancel scheduled emails**, each only if the app asked for it.
-   **Its limits**: who it can send to, which domains and From addresses, how many emails an hour and a day, whether each send waits for approval, and when the connection ends (7, 30 or 90 days).

Everything starts at the safest setting: read email details only, test addresses only, 20 emails an hour and 50 a day, approval required for every send, and 30 days. Widen only what you need. Allowing an app needs a fresh confirmation (your 2FA code or an emailed code).

The app acts as you, never with more than your role in the team allows. It can’t manage domains, API keys, webhooks, members, billing or settings, and it can’t send attachments. If you leave the team, it’s disconnected at once. [Using Vaartix with AI agents safely](https://vaartix.com/docs/guides/mcp-safety/) explains every control.

## See and disconnect apps

-   **Settings** → **Connected apps** lists every app connected to the team: who connected it, what it can do, its limits, when it was last used and how many calls it made today. Owners and admins can change its limits, pause it or disconnect it, or disconnect every app at once; members can disconnect their own.
-   **Logs** → **AI agent activity** shows every call an app made: the tool, whether it worked, how many recipients and which emails it created. Never the content.
-   **Settings** → **Audit log** records each app connected, changed and disconnected.

Access lasts an hour and is renewed automatically while you use the app. If a renewal token is ever used twice, which can mean it was stolen, Vaartix disconnects the app and the audit log says so.

## Use an API key instead

For an app that can’t sign in with OAuth, send an API key as the bearer token. The key’s own permissions and [limits](https://vaartix.com/docs/guides/mcp/#create-an-api-key) apply: a send-only key can send but can’t read. **Allow AI agents (MCP)** must be on for the team. Create the key as an **AI agent key** so it starts with safe limits.

```sh
claude mcp add --transport http vaartix https://mcp.vaartix.com \
  --header "Authorization: Bearer vx_live_your_key"
```

Keep the key out of shared or committed config files. Signing in is safer: there’s no long-lived secret, and you can disconnect the app in one click.

## Search the docs too

The [docs MCP server](https://vaartix.com/docs/guides/mcp/#docs-server) at `https://vaartix.com/docs/mcp` lets your assistant search and read these docs. It needs no sign-in.