集成
AgentSend works with any AI agent framework. Use the REST API directly or follow one of the framework-specific guides below.
概述
AgentSend exposes a simple REST API, so it fits naturally into any agent stack. Whether you're building with a high-level orchestration framework or calling the API by hand, the same primitives apply: create an inbox, send a message, read replies.
All integrations use the same underlying REST API. Framework-specific guides just show you how to wire it in idiomatically for each tool.
框架集成
选择您的框架获取分步指南,包括工具定义、示例智能体和 webhook 处理。
直接 API
对于上面未列出的任何框架,直接调用 REST API。每种有 HTTP 客户端的语言都可以直接使用。
await fetch(`https://api.agentsend.io/inboxes/${inboxId}/messages`, { method: "POST", headers: { "x-api-key": process.env.AGENTSEND_API_KEY, "Content-Type": "application/json", }, body: JSON.stringify({ to: ["user@example.com"], subject: "Hello from my agent", bodyText: "This was sent by an AI agent via AgentSend.", }), });
curl -X POST https://api.agentsend.io/inboxes/$INBOX_ID/messages \ -H "x-api-key: $AGENTSEND_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "to": ["user@example.com"], "subject": "Hello from my agent", "bodyText": "This was sent by an AI agent via AgentSend." }'
See the API Reference for the full list of endpoints, request schemas, and response examples.
Claude Code 与 Cursor
AgentSend exposes a /skill endpoint for MCP tool integration. Add it to your mcp_servers config and Claude Code or Cursor will automatically discover send_email, list_messages, and reply_to_thread as callable tools.
{
"mcpServers": {
"agentsend": {
"command": "npx",
"args": ["agentsend-mcp"],
"env": {
"AGENTSEND_API_KEY": "your-api-key",
"AGENTSEND_INBOX_ID": "your-inbox-id"
}
}
}
}For a full walkthrough of the MCP setup, see the Claude MCP integration guide.
社区
在未列出的框架中使用 AgentSend?我们很乐意看到。在我们的 Discord 上分享您的集成,我们会将其收录到文档中。