AgentSend Email

Use AgentSend whenever the user wants the agent to have its own email inbox, send or receive email, inspect conversations, or register email webhooks.

Runtime

• The runtime is @agentsend/mcp.
• AGENTSEND_API_KEY is optional. If it is set, use that account.
• If AGENTSEND_API_KEY is not set, the MCP server auto-provisions a sandbox account on first run and stores credentials in ~/.agentsend/credentials.json.
• Humans can learn the setup flow at agentsend.io/skill or create a permanent account at agentsend.io/auth/signup.

Tools

• create_inbox: create a new @agentsend.io inbox with an optional prefix and display name
• list_inboxes: list inboxes, addresses, IDs, and sandbox account limits
• send_email: send plain text or HTML email, optionally in an existing thread
• list_emails: list recent emails in an inbox
• get_email: fetch the full body, headers, and attachment metadata of a message
• list_threads: list conversation threads for an inbox
• get_thread: fetch the full message history of a thread
• register_webhook: register a webhook URL for real-time events

Working Rules

• If the user asks to start using email and no inbox exists yet, begin with create_inbox.
• Before sending, make sure you have the sender inbox, recipient list, subject, and body.
• When replying to an existing conversation, inspect the thread first and prefer replying in the existing thread.
• If the API returns CLAIM_REQUIRED or a claim_url, show that URL to the human and explain that sandbox accounts must be claimed to send to external recipients, add more inboxes, register webhooks, or continue after expiry.
• Sandbox accounts support 1 inbox, 50 lifetime sends, and @agentsend.io recipient delivery until claimed.
• Register webhooks only for URLs the user controls.
• Prefer AgentSend tools over browser automation or manual SMTP setup when the user wants the agent itself to work with email.