MCP server
Let AI agents such as Claude, Cursor and ChatGPT send email, check delivery and manage domains in your uSend account.
uSend runs a Model Context Protocol server, so an AI agent can work with your account: send an email, find out why one bounced, add a domain and read its DNS records, or read the replies you received.
| URL | https://mcp.usend.et/mcp |
| Transport | Streamable HTTP |
| Sign-in | OAuth (the agent opens uSend for you to approve it), or an API key |
Connect an agent
The quickest way is in the dashboard: Settings → Integrations → MCP server → Configure MCP shows the server URL and copy-ready setup for each tool below.
Claude
In Claude, open Settings → Connectors → Add custom connector, name it uSend and enter https://mcp.usend.et/mcp. Claude opens uSend in your browser: choose the organization and approve the access the agent asks for.
In Claude Code:
claude mcp add --transport http usend https://mcp.usend.et/mcpCursor and VS Code
Add the server to .cursor/mcp.json (Cursor) or .vscode/mcp.json (VS Code):
{
"mcpServers": {
"usend": { "url": "https://mcp.usend.et/mcp" }
}
}The editor signs you in through your browser the first time.
With an API key
Agents that cannot sign in through a browser (scripts, CI, self-hosted agents) send an API key instead:
{
"mcpServers": {
"usend": {
"url": "https://mcp.usend.et/mcp",
"headers": { "Authorization": "Bearer us_xxxxxxxxx" }
}
}
}A key with Sending access can only use send_email and send_batch; the other tools need Full access.
Tools
| Tool | What it does |
|---|---|
send_email | Send one email, or schedule it |
send_batch | Send up to 100 different emails |
get_email | An email's status, recipients, content, opens and clicks |
list_emails | Sent emails, newest first |
get_email_events | One email's delivery timeline, including bounce reasons |
cancel_email | Cancel a scheduled email |
list_domains | Your domains and whether each is verified |
get_domain | A domain's status and the DNS records to publish |
add_domain | Add a domain; returns its DNS records |
verify_domain | Check a domain's DNS records now |
list_received_emails | Email received on your domains |
get_received_email | One received email with its text and HTML |
list_logs | Recent API requests, to see why one failed |
Nothing the agent can do deletes data. Sending tools are marked so your agent asks before it sends.
Permissions and safety
- Every tool call is an ordinary API request made with your credential: the same limits, roles and rules apply, and it appears on the Logs page as
uSend MCP. - An OAuth connection follows the role of the person who approved it. If they are removed from the organization or lose the role, the agent stops working at once.
- Remove a connection under Settings → Team → Connected apps, or delete the API key.