Postmark API + MCP (2026): First SMTP Provider Shipping Official MCP

Postmark’s Email API v1 has been stable for over a decade, ships 13 official client libraries (more than any competitor in this review), and is the reference implementation for teams that want SMTP relay + rich webhooks + clean logs in a single tightly-scoped surface. Where it differentiates in 2026: Postmark is the first mainstream transactional email provider to ship an official Model Context Protocol server, letting Claude Desktop and Cursor send emails, list templates, and pull delivery stats directly through natural language. This tab inventories the API surface, the MCP capabilities, and the gotchas a buyer needs to evaluate before committing.

At a glance

MCP integration in 2026

Model Context Protocol is the emerging standard for connecting LLM agents to external tools. If you want an AI agent (Claude Desktop, ChatGPT, Cursor, custom OpenAI Agents SDK deployment) to send emails through Postmark, MCP is now the connection layer of choice. Here is where Postmark stands.

Available MCP servers for Postmark

Three ways to reach Postmark from an AI agent as of August 2026, ordered by maturity. The first is official Postmark. The others are bridges through general-purpose MCP layers.

Email API v1 essentials

The Postmark Email API follows REST conventions with minimal quirks. Its scope is intentionally narrow: transactional email send and receive, plus supporting resources (templates, sender signatures, webhooks, bounces).

Endpoints are grouped by concern (email, bounces, templates, servers, message streams, domains, sender signatures, statistics, webhooks, suppressions, data removal, inbound rules triggers) with no deep resource nesting. All requests are stateless: authenticate every call with the appropriate token header. No cursor pagination beyond simple count + offset query parameters where lists are returned.

Authentication methods

Two token types, both operating through custom HTTP headers.

Server Token (X-Postmark-Server-Token)

Used for the majority of API calls: sending email, managing templates, reading messages, bounces, and stats for a single Server. Scoped to one Postmark Server (a Server in Postmark parlance is a container for domains, message streams, and tokens). Accessible to Account Owners, Admins, and Server Admins in the Postmark UI. Copy from Server › API Tokens.

Account Token (X-Postmark-Account-Token)

Used for account-level operations: creating new Servers, managing sender signatures, listing domains across all Servers, initiating data removal requests, managing account-wide DMARC monitoring. Restricted to Account Owners and Admins. Copy from Account › API Tokens. Never expose an Account Token to browser code or client-side JavaScript.

Rate limits

Postmark’s rate-limiting posture is unusually permissive for a modern API: no daily quota, no per-endpoint budget, one loosely-enforced concurrency ceiling.

Official SDKs

Postmark ships one of the deepest official SDK catalogs in transactional email. Thirteen libraries maintained by ActiveCampaign, covering every mainstream language plus specialized adapters for Rails, WordPress, Craft CMS, and Grunt build pipelines.

Endpoints reference

The endpoint groups you will actually use in production. Full reference at postmarkapp.com/developer/api covers roughly 100 individual operations across all groups.

Code examples

Two representative snippets covering the most common first-day tasks: send a single email, then send a batch with a template.

Node.js: send a single email

const postmark = require('postmark');

const client = new postmark.ServerClient('YOUR_SERVER_API_TOKEN');

client.sendEmail({
  From: 'sender@yourdomain.com',
  To: 'recipient@example.com',
  Subject: 'Test from Postmark API',
  HtmlBody: '<h1>Hello</h1><p>This is a test.</p>',
  TextBody: 'Hello. This is a test.',
  MessageStream: 'outbound'
}).then(response => {
  console.log(`Sent: ${response.MessageID}`);
}).catch(err => {
  console.error(`Error: ${err.message}`);
});

Python: send a batch with a template

from postmark import PMBatchMail

messages = [
    {
        'From': 'sender@yourdomain.com',
        'To': recipient,
        'TemplateAlias': 'welcome-email',
        'TemplateModel': {
            'name': name,
            'product_url': 'https://yourdomain.com/dashboard'
        },
        'MessageStream': 'outbound'
    }
    for (recipient, name) in [
        ('alice@example.com', 'Alice'),
        ('bob@example.com', 'Bob'),
        # ... up to 500 messages per batch
    ]
]

batch = PMBatchMail(api_key='YOUR_SERVER_API_TOKEN', messages=messages)
result = batch.send()
print(f'Batch sent: {len(result)} messages processed')

Common gotchas

Deprecations and changelog

Postmark maintains a public changelog and release notes on the Updates page. Key items to know:

• Email API v1 is current. No successor announced. No breaking version bumps since 2010 — unusual stability for an API of this age.
• Postmark MCP Server released June 2025 (Postmark Labs) as the first official MCP for transactional email. Expanded scope expected as MCP HTTP transport matures per the 2026-07-28 spec.
• SDK maintenance cadence: Node.js library sees regular releases. Other language SDKs update as needed — no frozen SDKs equivalent to Mailchimp’s 2022 pause.
• Message Streams generally available since 2020. Older accounts still have a legacy single-stream setup by default — migrate to streams for reputation isolation between transactional and broadcast traffic.
• DMARC monitoring add-on released 2024 in response to Gmail/Yahoo/Microsoft bulk-sender rules. $14/mo per domain, parses aggregate reports into a dashboard.

Frequently asked questions