AI Agents & Automations

Any agent framework or automation tool that can make an authenticated HTTP request can send through SendAPI. Connect over MCP for typed tools, or call the REST API directly. The pattern is always the same:

• Base URL: https://sendapi.co/v1
• Auth header: Authorization: Bearer sk_live_xxxx
• Body: JSON. Responses are wrapped as { "data": … }; errors as { "error": { "message", "code" } }.

Claude, ChatGPT, Cursor

Connect the MCP server and the agent gets every SendAPI tool instantly — the fastest path, with no code. See the MCP page for the one‑line Claude Code command and the Cursor/ChatGPT connector config.

n8n

Two options:

• AI Agent (MCP): add an MCP Client Tool node pointed at https://sendapi.co/mcp with an Authorization: Bearer sk_live_xxxx header. Your n8n AI Agent can then call any SendAPI tool.
• No agent (HTTP Request): add an HTTP Request node, create a Header Auth credential (Header Name Authorization, Value Bearer sk_live_xxxx), and POST to any endpoint:

{
  "to": "+15551234567",
  "content": "Hello from n8n"
}

Zapier & no‑code

Zapier doesn't run MCP servers, but you can call SendAPI in any Zap:

1. Add an action step → Webhooks by Zapier → Custom Request.
2. Method POST, URL https://sendapi.co/v1/sms/send (or any endpoint).
3. Data (JSON): { "to": "+15551234567", "content": "Hello from Zapier" }
4. Headers: Authorization: Bearer sk_live_xxxx and Content-Type: application/json.

The same recipe works in Make, Pipedream, Retool, or any tool with an HTTP step.

Custom agents

Building your own agent? Call the REST API directly, or use an official SDK:

npm install @sendapi/node     # Node / TypeScript
pip install sendapi           # Python
composer require sendapi/sendapi  # PHP

The complete request and response contract for every endpoint lives in the OpenAPI spec and the API Reference.

Requirements

• A SendAPI API key, sent on every request as a Bearer token.
• An MCP‑capable client or any tool that can send an authenticated HTTP POST.
• For WhatsApp, a connected session; for production email, a verified sender domain.

Safety

Agents send real messages that cost money and reach real people. Always confirm the recipient and content before sending, and require explicit confirmation before any bulk send (up to 500 recipients per call). Check remaining quota with get_usage first.

Related

• MCP Server — typed tools for any MCP client.
• Agent Skills — best‑practice know‑how for Claude.