How to Connect Postmark to Claude (and What It Can't Do)
Postmark is further ahead on Claude support than almost any transactional email service. ActiveCampaign, Postmark’s parent company, ships an official open-source MCP server with 24 tools covering sending, templates, message search, delivery diagnostics, bounces, suppressions, stats, and webhooks. What Postmark still isn’t, as of mid-2026, is a one-click app in Claude’s connector directory — you install the server yourself, and Claude’s custom-connector features sit behind a paid Claude plan. The other constant: Claude only calls those tools during a conversation you’re actively having. A bounce webhook firing at 2am does nothing until you open a chat and ask about it.
What the official server covers
The npm package is @activecampaign/postmark-mcp, typically run locally via npx and registered in Claude Desktop’s MCP configuration. Its 24 tools break down like this:
- Sending (write):
sendEmail,sendEmailWithTemplate, plus batch variants — real sends through your server token and message stream, not drafts. - Templates (read + write): list, fetch, create, edit, delete, and
validateTemplate, which checks a template’s model against its variables before anything goes out. - Investigation (read-only):
searchOutboundMessages,getMessageDetails, anddiagnoseDeliveryfor “why didn’t this receipt arrive” archaeology. - Bounces and suppressions: search bounces, pull the raw bounce dump, reactivate a bounced address, and create or delete suppressions.
- Stats and webhooks: delivery stats by stream, server info, and webhook CRUD (HTTPS-only, with an optional URL allowlist).
Configuration is three environment variables — POSTMARK_SERVER_TOKEN, DEFAULT_SENDER_EMAIL, DEFAULT_MESSAGE_STREAM — and the token never leaves your machine; the server talks straight to Postmark’s API.
Conversations that work well against this tool set:
- “Diagnose why this morning’s receipts to yahoo.com addresses keep soft-bouncing on the outbound stream.”
- “Validate my invoice template and list every variable its model expects.”
- “Search outbound messages to that customer, find the hard bounce, reactivate it, and resend the receipt.”
One security note Postmark users should not skip
In September 2025, an attacker published an unscoped npm package called postmark-mcp that behaved normally for fifteen versions, then added a one-line backdoor BCC-ing every email to an attacker-controlled address. Postmark had nothing to do with it, but roughly 1,500 downloads happened before removal. The lesson is narrow and practical: install only the scoped @activecampaign/postmark-mcp package, and follow the repo’s own advice — give the MCP its own dedicated Postmark server so the token’s blast radius stays small, and never auto-approve send or delete tools. Well-behaved MCP clients, Claude Desktop included, auto-approve only the read-only tools and prompt you before anything that writes.
The boundary: everything waits for you
Postmark’s whole value is event-shaped — bounce webhooks, spam-complaint webhooks, delivery events per stream. Claude’s connector model inverts that. The MCP can create a webhook for you, but Claude can’t receive one: there is no mechanism by which an inbound event starts a Claude conversation or resumes an old one. So the flows a transactional-email owner actually wants — reactivate-and-resend when a password-reset email hard-bounces, an alert when the complaint rate on the broadcast stream crosses 0.1%, a morning digest of yesterday’s delivery stats — all stall at the same point. Claude executes each of them beautifully on request, and only on request.
Closing the loop with Carly
Carly is an AI executive assistant built for exactly the event-shaped half. It runs in the cloud and fires on triggers, so the Postmark workflows above stop needing you in the middle: a hard bounce on a magic-link email can kick off a resend and a Linear issue; a complaint-rate spike can page you in Slack with the offending campaign attached; delivery stats can land in your inbox every morning at 8 without anyone prompting.
You describe the outcome in plain English — “when a receipt bounces, retry once, then email the customer from my Gmail asking for a better address” — and Carly interviews you about edge cases before building the workflow with you. There’s no MCP server to keep patched and no token config to babysit. AI agents start at $35/month, and steps in a workflow that don’t use AI run free and unlimited. Carly connects to 200+ tools — see integrations and the Postmark integration page.
Official Postmark MCP vs Carly
| Claude + Postmark MCP | Carly | |
|---|---|---|
| Send via templates & batch | Yes, in chat | Yes, in workflows |
| Search messages & diagnose delivery | Yes, in chat | Yes |
| React when a bounce webhook fires | No — can’t receive events | Yes |
| Morning delivery-stats digest, unprompted | No | Yes |
| Manage suppressions on a schedule | No — manual asks only | Yes |
| Server maintained by | You (npx, env vars, updates) | Carly (cloud) |
| Requires | Paid Claude plan + server token setup | AI agents from $35/mo |
Frequently Asked Questions
Does Claude integrate with Postmark?
Yes, better than most email tools. ActiveCampaign publishes an official Postmark MCP server (@activecampaign/postmark-mcp) with 24 tools for sending, templates, bounces, suppressions, diagnostics, and webhooks. It’s not a one-click directory app as of mid-2026 — you run it yourself and configure it in Claude, and custom connectors require a paid Claude plan.
Is the Postmark MCP server safe to use?
The official scoped package is open source, MIT-licensed, and keeps your server token local. The caution flag is the 2025 incident where a fake unscoped postmark-mcp npm package BCC’d users’ emails to an attacker. Install only @activecampaign/postmark-mcp, use a dedicated Postmark server for it, and keep send/delete tools behind manual approval.
Can Claude resend an email when it bounces?
On command, yes — it can find the bounce, reactivate the address, and resend, all in one conversation. Automatically, no: Claude can’t receive Postmark’s bounce webhooks or act between chats. Trigger-driven resends need an agent platform like Carly.
What do I need before connecting?
A Postmark server token, a verified sender signature, a message stream name, Node available for npx, and a paid Claude plan for the connector side. The repo’s README covers the Claude Desktop config block.
How does Carly handle Postmark differently?
Carly listens for events instead of waiting for prompts: bounces, complaint spikes, and finished sends can each start a workflow that emails, files, alerts, or updates other tools. AI agents start at $35/month.
More: Claude connectors · Can Claude send emails · Claude vs Carly · Claude Cowork alternatives · Best AI email tools · Claude + Resend · Claude + SendGrid · Claude + Mixmax
Ready to automate your busywork?
Carly schedules, researches, and briefs you—so you can focus on what matters.
See what people say
"Before Carly, I relied on a Calendly link, but the whole process felt impersonal and not very professional. Carly changed that by handling all the back-and-forth, so I'm no longer stuck in endless email threads trying to line up schedules.
Now Carly reaches out to candidates, shares my real-time availability, lets them pick a slot, then sends a Zoom link and drops it straight into my calendar. She sends reminders to both of us before each call, which has significantly reduced no-shows and last-minute confusion.
On top of scheduling, Carly acts like a full executive assistant, sending me my schedule the night before so I can prepare for each call. It reminds me of the old x.ai assistant, but Carly is noticeably smarter, faster, and better suited to my healthcare recruitment business."
