MCP

What is MCP (Model Context Protocol)? AI Tool Integration Explained (2026)

Q: How does the Sendblue MCP server work?

The Sendblue MCP server (package: sendblue-mcp, install command: npx -y sendblue-api-mcp) exposes the Sendblue iMessage API as MCP tools. Once configured in your AI client, the AI can call tools like send_message, get_messages, and lookup_number to send and receive iMessages as part of any agent workflow. It requires a Sendblue API key.

MCP is the open standard that lets AI models connect to real-world tools. Sendblue's MCP server adds iMessage to any AI workflow — one install command away.

What is MCP?

MCP (Model Context Protocol) is an open standard created by Anthropic that defines how AI models connect to external tools, data sources, and services. Released in late 2024 and rapidly adopted across the AI ecosystem, MCP solves a fundamental problem: AI models are powerful reasoners but isolated — they cannot take action in the real world without a standardized way to call external systems.

MCP acts like a plugin system for AI. Developers publish MCP servers that expose capabilities (called tools, resources, and prompts). AI clients — like Claude Code, Cursor, and Windsurf — connect to those servers and can call their tools during a conversation or agentic task.

Think of MCP as the USB standard for AI integrations: one protocol, any tool, any compatible AI client.

How MCP Works

AI Client
(Claude Code, Cursor, Windsurf)

↔

MCP Server
(sendblue-mcp)

→

Sendblue API
(iMessage delivery)

→

iPhone Recipient
(blue bubble)

The AI client calls the MCP server's tools. The server calls Sendblue's API. The message arrives as a blue-bubble iMessage.

The MCP protocol uses a client-server architecture over standard I/O or HTTP/SSE:

The AI client discovers available tools by querying the MCP server's tool manifest
During a task, the AI decides to call a tool (e.g., send_message)
The client sends a JSON-RPC request to the MCP server with the tool name and arguments
The MCP server executes the action (calls the Sendblue API) and returns the result
The AI incorporates the result and continues the task

Why MCP Matters

AI agents take real action

Without MCP, AI models can only generate text. With MCP, they can send messages, query databases, create CRM records, and interact with APIs — turning conversations into completed tasks.

One integration, all clients

Before MCP, every AI tool required custom integrations. Now, publish one MCP server and it works in Claude Code, Cursor, Windsurf, VS Code, and any other MCP-compatible client.

Open standard

MCP is open source and not proprietary to Anthropic. Any developer can build MCP servers and clients. The ecosystem is growing rapidly across the AI tooling landscape.

Secure and scoped

MCP servers define exactly which tools they expose. Clients must explicitly call each tool. This keeps AI actions auditable and bounded.

The MCP Ecosystem in 2026

MCP clients (tools that can use MCP servers) include:

Claude Code — Anthropic's official CLI for Claude, the reference MCP client
Claude Desktop — The desktop application for Claude with MCP server support
Cursor — AI-powered code editor with MCP tool calling
Windsurf — Codeium's AI IDE with MCP support
VS Code — Via Copilot and third-party MCP extensions
LangChain / LlamaIndex — Agent frameworks with MCP adapters

The MCP server registry (published at modelcontextprotocol.io) lists hundreds of servers covering databases, file systems, web search, CRM systems, payment processors, and communication tools — including Sendblue.

Sendblue MCP Server: iMessage for AI

Sendblue publishes an official MCP server — sendblue-mcp — that exposes the Sendblue iMessage API as MCP tools. Once configured in your AI client, the AI can send and receive iMessages as part of any agentic workflow.

Install in one command

npx -y sendblue-api-mcp

Configure in your AI client

// Add to your MCP client config (e.g., Claude Code .mcp.json)
{
  "mcpServers": {
    "sendblue": {
      "command": "npx",
      "args": ["-y", "sendblue-api-mcp"],
      "env": {
        "SENDBLUE_API_KEY_ID": "your_api_key_id",
        "SENDBLUE_API_SECRET_KEY": "your_secret_key"
      }
    }
  }
}

What the AI can now do

With the Sendblue MCP server configured, your AI agent can:

Send iMessages to any phone number as blue bubbles
Look up whether a number has iMessage enabled
Retrieve message history for a conversation
Send media attachments (photos, videos) via iMessage
Trigger iMessage outreach as part of multi-step agent workflows

Example AI workflow

// Prompt to Claude Code with Sendblue MCP configured:
"Check our CRM for leads who signed up today but haven't
been contacted. Send each of them an iMessage introducing
our product. Use their first name."

// Claude will:
// 1. Query the CRM (via another MCP tool or API)
// 2. For each new lead, call sendblue.send_message
// 3. Messages arrive as blue bubbles on recipients' iPhones

Why iMessage + AI Is Powerful

iMessage gets 30–45% response rates compared to 5–10% for SMS. When AI agents use iMessage to reach out, they are not just automating messages — they are automating engagement at the highest-response channel available.

Combined with AI's ability to personalize at scale, read responses, and adjust follow-up timing, iMessage-powered AI agents outperform traditional SMS automation significantly. Learn more at the Sendblue MCP page or the iMessage API glossary page.

Frequently Asked Questions

What is MCP (Model Context Protocol)?

MCP (Model Context Protocol) is an open standard created by Anthropic that defines how AI models connect to external tools, data sources, and services. It works like a plugin system for AI: developers publish MCP servers that expose capabilities (tools, resources, prompts), and AI clients like Claude Code, Cursor, and Windsurf can connect to those servers to call real-world functionality during a conversation or task.

How does the Sendblue MCP server work?

The Sendblue MCP server (package: sendblue-mcp, install command: npx -y sendblue-api-mcp) exposes the Sendblue iMessage API as MCP tools. Once configured in your AI client, the AI can call tools like send_message, get_messages, and lookup_number to send and receive iMessages as part of any agent workflow. It requires a Sendblue API key.

Which AI tools support MCP?

MCP is supported by Claude Code (Anthropic's CLI), Claude Desktop, Cursor, Windsurf, VS Code (via extensions), and a growing ecosystem of AI agent frameworks. Any MCP-compatible client can connect to the Sendblue MCP server and gain iMessage capabilities without writing custom API integration code.

What can I build with the Sendblue MCP server?

With the Sendblue MCP server you can build AI agents that: automatically send iMessage follow-ups to new leads, respond to inbound iMessages using AI, trigger iMessage outreach as part of multi-step agent workflows, and integrate iMessage into any automation that runs inside an MCP-compatible AI tool. Because iMessage gets 30–45% response rates vs 5–10% for SMS, AI-powered iMessage outreach is significantly more effective than SMS-based alternatives.

Add iMessage to your AI workflows

Install the Sendblue MCP server and let Claude, Cursor, or any AI tool send blue bubbles.

View MCP Docs