MCP Server for AI Agents

zend.sh includes a built-in Model Context Protocol (MCP) server that lets AI agents manage your email campaigns through natural language. Connect Claude, GPT, or any MCP-compatible agent to create campaigns, enroll leads, and check analytics.

Endpoint

POST https://app.zend.sh/api/mcp

Transport: Streamable HTTP (application/json responses)

Authentication

The MCP server uses the same API key authentication as the REST API:

Authorization: Bearer zk_live_...

Available tools

Tool	Description
`list_accounts`	List connected mailbox accounts with health status
`create_campaign`	Create a new campaign with name, send window, and settings
`list_campaigns`	List all campaigns with status and stats
`start_campaign`	Start a campaign (must be in draft state with steps and leads)
`enroll_leads`	Add leads to an existing campaign
`get_analytics`	Get team-wide or per-campaign analytics

Configure with Claude Desktop

Add the zend.sh MCP server to your Claude Desktop configuration at ~/.claude/claude_desktop_config.json:

{
  "mcpServers": {
    "zend": {
      "url": "https://app.zend.sh/api/mcp",
      "headers": {
        "Authorization": "Bearer zk_live_..."
      }
    }
  }
}

Restart Claude Desktop after updating the configuration.

Example conversations

Once connected, you can manage campaigns through natural language:

Creating a campaign:

"Create a campaign called 'Q1 SaaS Outreach' that sends Monday through Friday, 9 AM to 5 PM Eastern. Stop sending to a lead if they reply."

Enrolling leads:

"Add these leads to the Q1 SaaS Outreach campaign: jane@acme.com (Jane, Acme Corp), bob@globex.com (Bob, Globex Inc)"

Checking analytics:

"How is the Q1 SaaS Outreach campaign performing? Show me sent, delivered, and reply rates."

Managing accounts:

"List my connected mailbox accounts and their health scores."

Programmatic usage

You can also connect to the MCP server programmatically using the MCP SDK:

import { Client } from '@modelcontextprotocol/sdk/client/index.js'
import { StreamableHTTPClientTransport } from '@modelcontextprotocol/sdk/client/streamableHttp.js'

const transport = new StreamableHTTPClientTransport(
  new URL('https://app.zend.sh/api/mcp'),
  {
    requestInit: {
      headers: {
        Authorization: 'Bearer zk_live_...',
      },
    },
  }
)

const client = new Client({ name: 'my-app', version: '1.0.0' })
await client.connect(transport)

// List available tools
const tools = await client.listTools()

// Call a tool
const result = await client.callTool({
  name: 'get_analytics',
  arguments: {},
})

Security

The MCP server enforces the same authentication and rate limits as the REST API
All operations are scoped to the team associated with the API key
Tool inputs are validated with the same schemas as the REST endpoints
The server uses JSON responses (not SSE streams) for compatibility with serverless environments