MCP server

Point Claude, ChatGPT, or Cursor at your sending data.

Sendfeather speaks the Model Context Protocol, so an AI assistant can send mail and answer delivery questions — what went out, what bounced, how each mailbox provider is treating the mail — from inside the chat.

Connect your assistant

Create an API key under API keys in the dashboard. Reads need the read:activity scope; sending and do-not-send changes need the send scope. Then add the server to your MCP client:

json
{
  "mcpServers": {
    "sendfeather": {
      "type": "http",
      "url": "https://sendfeather.com/api/mcp",
      "headers": { "Authorization": "Bearer ulk_live_..." }
    }
  }
}

What it can do

  • send_email — send through the same pipeline as the API, from a verified domain
  • get_delivery_stats — delivery, open, click, bounce, and complaint counts and rates for a window of days
  • get_deliverability_by_provider — the Gmail, Outlook, Yahoo, and Apple Mail breakdown
  • list_emails and get_email — recent mail and one message with its delivery events
  • list_suppressions, add_suppression, remove_suppression — curate the do-not-send list
  • list_domains — verification state and outstanding DNS records

Every call runs through the same validation, rate limits, and suppression rules as the REST API, and the key scopes it to one project.