SenderKit

SenderKit is the messaging layer for AI agents and applications. One API sends transactional email, SMS, push, and web-push, and receives inbound mail back — so anything you build, from a background job to an autonomous agent, can reach a user and act on what comes back.

This connector puts that infrastructure directly in Claude's hands. Claude sends real messages on your workspace, writes and refines templates, and inspects what happened afterwards — then wires the same calls into your application when you're ready to ship.

What Claude can do:

• Send transactional & marketing messages across email, SMS, push, and web-push — from a saved template or with inline content
• Create, inspect, and regenerate templates from a plain-language brief, on

any channel

• Provision inbound email addresses and custom inbound domains, then read the

mail that arrives — so agents can receive as well as send

• List sent and scheduled messages, filter by channel or status, and check

delivery detail for any individual message

• Cancel a scheduled or queued message before it goes out
• Confirm the connected workspace and whether sending is live or test
• Wire SenderKit into your own codebase — Claude can add the send calls,

template handling, and webhooks to your app

Typical uses: prototype a welcome email or OTP in conversation, then ship the same template call in code; debug a delivery that failed or stalled; audit what an agent sent last week; give a long-running agent an inbox it can read replies from.

SenderKit sends directly and can also route through providers you already run — Resend, SendGrid, Postmark, Mailgun, SES, SMTP, Twilio, FCM, APNs, Expo — so it fits in front of existing infrastructure rather than replacing it. Test mode simulates sends until you switch to live.