mcpdirectory.dev
9,129 servers · updated 23 Sept

Mailbuttons MCP server

official from Mailbuttons Active last commit 2 mo
{
  "mcpServers": {
    "mailbuttons": {
      "command": "npx",
      "args": [
        "-y",
        "@mailbuttons/mcp-server@latest"
      ],
      "env": {
        "MAILBUTTONS_API_KEY": "<mailbuttons_api_key>"
      }
    }
  }
}
file .mcp.json (in your project root)install npx -y @mailbuttons/mcp-serverversion 0.1.0

What it does

This MCP server provides governed email for AI agents with sandbox inboxes, policy enforcement on sending, and audit logging. It is for developers who need to give agents email access while maintaining security guardrails enforced server-side rather than through prompts.

  • Allow an agent to receive and read emails from a restricted set of approved senders without risk of prompt injection.
  • Enable an agent to compose and send emails while a policy gate blocks sends to recipients on a blocklist and requires human approval for external communication.
  • Audit all email activity including refused sends through a tamper-evident hash-chained log.

Tools this server exposes

ToolDoes
read_mailRetrieve incoming emails from the agent's sandbox mailbox.
send_mailSend an email; blocked recipients return a status result rather than an error.
propose_external_sendRequest approval from a named human to send externally or widen mailbox scope.
check_recipient_policyRead the blocklist for the mailbox without modifying it.
view_audit_logAccess the tamper-evident audit log recording all sends and refusals.

Before you enable it

An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.