Menu
Mattering Automations connector

A Postmark MCP that runs the whole account, not just the send

87 nodes across the fifteen sections of Postmark's REST API. Transactional and template sends, bulk broadcast, message search and raw MIME, a sending overview and eleven daily statistics breakdowns, bounces and suppressions, message streams, webhooks and inbound rules, and the account-level work most integrations skip: servers, domains and sender signatures. Both of Postmark's tokens are handled, and a bulk send cannot leave on a transactional stream.

87 ready-made nodes One credential to connect Works with Claude, ChatGPT, and Matter

What the Postmark MCP connector does

The Postmark connector for Mattering Automations is 87 nodes on Postmark's REST API at api.postmarkapp.com, which is one global host with no regional variants. Sending covers a single message, a batch of up to 500, a template render, a template batch, and an asynchronous bulk broadcast with its own request status. Reading covers outbound and inbound message search, full message detail with the delivery event timeline, raw MIME source, and every recorded open and click. Reporting covers a sending overview plus eleven daily breakdowns: sent, bounced, spam complaints, tracked, opens, clicks, platform and client usage, and where in a message the link was clicked. Operations covers bounces and reactivation, per-stream suppressions, message streams, webhooks and inbound blocking rules. Administration covers servers, sending domains and sender signatures, template push between servers, and data removal. Forty four nodes read and forty three act.

Every capability is a node on the Automations canvas, and every node speaks MCP, the open protocol for connecting AI assistants to tools. Claude, ChatGPT, and any other MCP-capable assistant can use them, alongside Matter, the AI assistant built into the platform. The same node is also a step in a visual workflow, which is the part an assistant on its own cannot give you: a deliverability check that runs at 7am on a schedule with nobody present, chained to whatever else it needs, holding one credential in a shared workspace rather than in one person's local config.

Postmark does publish its own MCP server. It is at github.com/ActiveCampaign/postmark-mcp, maintained under Postmark's parent company, and linked from Postmark's developer page under AI tooling. It exposes 24 tools covering sending, templates, message search, bounces, suppressions, webhooks, delivery stats and server info, and if that is the whole of what you need from an assistant it is a perfectly good answer. The differences worth knowing are these, and each is checkable rather than rhetorical. That server is configured with a Postmark server token, a sender address and a message stream, so the account-level surface, which is servers, domains, sender signatures, template push and data removal, is outside its scope and inside this connector's. The daily statistics breakdowns, message stream management, bulk broadcast, inbound message handling and inbound rules are likewise nodes here. And a node runs unattended: on a schedule or a trigger, chained with every other connector in one workflow, with the credential held once for a workspace and destructive operations flagged as destructive.

Postmark authenticates with two different tokens and this is the single most common way a Postmark integration goes wrong. A server token, sent as X-Postmark-Server-Token, covers sending, messages, statistics, bounces, templates, message streams, suppressions, webhooks, inbound rules and the server behind the credential, which is 62 of the 87 nodes. An account token, sent as X-Postmark-Account-Token, covers servers, domains, sender signatures, pushing templates between servers and data removal, which is the other 25. They are not interchangeable, and they live on different pages of the Postmark interface. The account token field on the credential is optional, so the 62 server-level nodes work with nothing but the server token, and the 25 account-level nodes stop before the request is made with a message that names the token and the page rather than handing you a bare 401 to interpret.

Message streams are the heart of the product rather than a setting, and the connector treats them that way. Postmark keeps Transactional, Broadcast and Inbound as separate streams precisely so that bulk mail cannot damage the deliverability of receipts and password resets, and Postmark's own send endpoint defaults to the reserved transactional stream, named outbound, when no stream is given. So the bulk node's stream picker is pinned to Broadcast streams and simply does not offer a transactional one; the reserved outbound id is refused by name before any network call; and the chosen stream is read back and rejected unless Postmark reports its type as Broadcasts. There is deliberately no override flag. An escape hatch here would be the mistake wearing a checkbox, and the guard costs one cheap read on the one operation where an extra round trip is irrelevant.

Eight nodes are classed destructive and each says what it costs in its own description. Deleting a server removes its tokens, streams, templates, webhooks and message history. Deleting a domain stops every sender signature on it. Archiving a message stream stops sending immediately and Postmark deletes archived streams 45 days after archiving, restorable only until then. Deleting a webhook stops events with no replay, so anything firing in the gap is lost. Data removal is the sharpest of them: it erases a recipient's history and content for a subject request, Postmark documents no way to recover it, and Postmark grants access to the endpoint by request through support rather than enabling it by default.

You can try this without paying anything. Postmark starts new accounts on a free developer plan that offers 100 emails a month to test an integration with, which Postmark states never expires or runs out, and which includes the REST API, SMTP and Sandbox Mode. Sandbox Mode is the useful part for automation work: set a server's delivery type to Sandbox and it accepts mail and discards it, so the send nodes can be exercised without spending the allowance or reaching a real inbox. Leaving it on by accident is exactly as bad as it sounds, which is why the node that sets it says so.

Two behaviours of the API are worth knowing before you build a report on it. Postmark evaluates the fromdate and todate filters on statistics, messages, bounces and suppressions in United States Eastern time rather than UTC, so a window assembled from UTC dates is shifted by hours at both ends and quietly moves rows between day buckets. And message data expires at the end of your retention period, which Postmark documents as 45 days by default and configurable from 7 to 365 days, so a search that reaches past it returns nothing rather than an error. Both are stated on the fields they affect.

One connection, three surfaces

  • Visual automations. Drag Postmark nodes onto the canvas and wire them to triggers, conditions, and other apps.
  • AI agents, yours or ours. These nodes speak MCP, the open protocol, so Claude, ChatGPT, and any MCP-capable assistant can use them, and Matter, the agent built into the platform, uses them natively.
  • The whole platform. Combine with native tasks, projects, and other connectors, no middleware in between.

Every Postmark node, ready to drop on the canvas

87 nodes, generated from the connector itself so this list always matches what ships. Expand a category to browse.

Read and query nodes 44

Postmark: Get Bounce

Read one bounce in detail, including the receiving server's own explanation and whether the address can be reactivated.

Postmark: Get Bounce Counts

Daily bounce counts split by kind: hard bounces, soft bounces, transient delays and SMTP API errors. Days with no bounces of a kind simply omit that key.

Postmark: Get Bounce Source

Get the raw SMTP conversation for a bounce. Postmark keeps bounce dumps for 30 days, so older bounces return nothing.

Postmark: Get Bulk Request

Read one bulk send request's status, percentage complete, released count and failed count.

Postmark: Get Click Browser Usage

Daily clicks broken down by browser family. Browser names vary by account and by window, so the columns are not a fixed set.

Postmark: Get Click Counts

Daily link click counts, total and unique. As with opens, totals add across days and uniques do not.

Postmark: Get Click Location

Daily clicks split by where in the message the link was clicked, HTML part or text part.

Postmark: Get Click Platform Usage

Daily clicks split by platform: desktop, mobile and unknown.

Postmark: Get Clicks For Message

List every recorded link click for one specific message.

Postmark: Get Data Removal Status

Check whether a data removal request is still pending or has completed. A status of Done means the erasure has happened and cannot be undone. Needs the account token.

Postmark: Get Delivery Stats

Lifetime bounce totals by type for this server, plus the number of addresses currently deactivated because of bounces.

Postmark: Get Domain

Read one domain in full, including the exact DNS records to publish for DKIM and Return-Path. Needs the account token.

Postmark: Get Email Client Usage

Daily opens broken down by email client. Client names are whatever Postmark detected for your recipients, so the columns differ between accounts and between windows.

Postmark: Get Inbound Message

Read one received message in full, including bodies, every header, attachment metadata and, when it was blocked, the rule that blocked it.

Postmark: Get Message Stream

Read one message stream, including its type and its unsubscribe handling.

Postmark: Get Open Counts

Daily open counts, total and unique. Totals add up across days but uniques do not: one recipient opening on Monday and Tuesday is one unique over the week, not two.

Postmark: Get Open Platform Usage

Daily opens split by platform: desktop, mobile, webmail and unknown.

Postmark: Get Opens For Message

List every recorded open for one specific message.

Postmark: Get Outbound Message

Read one sent message in full, including its bodies, metadata and the whole event timeline of deliveries, transient failures, opens, clicks, bounces and subscription changes.

Postmark: Get Outbound Message Source

Get the raw MIME source of a sent message, headers and all. Useful when a recipient's mail server disputes what was actually delivered.

Postmark: Get Sender Signature

Read one sender signature, including its confirmation state and its DKIM and Return-Path records. Needs the account token.

Postmark: Get Sending Overview

One summary of sending health over a window: sent, bounced, spam complaints, opens, tracked mail and clicks, plus bounce and complaint rates. The rates are percentages of sent and must never be added across windows.

Postmark: Get Sent Counts

Daily counts of messages sent, with the window total alongside them.

Postmark: Get Server

Read the configuration of the server this credential belongs to, including its inbound address, webhook URLs and tracking defaults. API tokens are returned by Postmark and should be treated as secrets.

Postmark: Get Server By ID

Read any server on the account by ID. Needs the account token.

Postmark: Get Spam Complaint Counts

Daily counts of spam complaints. A rising line here is the earliest reliable warning that a sending reputation is in trouble.

Postmark: Get Template

Read one template's subject, HTML body, text body and layout by ID or alias.

Postmark: Get Tracked Email Counts

Daily counts of messages sent with tracking enabled. This is the denominator for any honest open or click rate.

Postmark: Get Webhook

Read one webhook's URL, authentication, custom headers and event triggers.

Postmark: Get Webhook Statistics

Delivery health for one webhook over Postmark's own fixed trailing 24 hours: request counts, success rate, retries and response times, overall and per trigger. The window is not selectable.

Postmark: List Bounces

List bounces on this server, filterable by bounce type, address, tag, message, stream and date window. The Inactive flag tells you whether Postmark has stopped sending to that address.

Postmark: List Bulk Requests

List recent bulk send requests on this server with their progress, released counts and failure counts.

Postmark: List Domains

List the sending domains on the account with their SPF, DKIM and Return-Path verification status. Needs the account token.

Postmark: List Inbound Rules

List the addresses and domains blocked from delivering inbound mail to this server.

Postmark: List Message Clicks

List link click events across the server, including which link was clicked and whether it was clicked in the HTML or the text part. Link tracking must be enabled for events to exist.

Postmark: List Message Opens

List open events across the server with the client, operating system, platform and approximate location recorded for each. Open tracking must be enabled for events to exist.

Postmark: List Message Streams

List the message streams on this server with their types. Transactional, Broadcast and Inbound are separate streams and the type decides what a stream may legitimately carry, so this is the list to check before sending anything.

Postmark: List Sender Signatures

List the confirmed and pending From addresses on the account. Needs the account token.

Postmark: List Servers

List every server on the Postmark account. Needs the account token, which is a different token from the server token used for sending.

Postmark: List Suppressions

List the suppressed addresses on one message stream, with why each was suppressed and who suppressed it. Suppressions are per stream, so an address blocked on your broadcast stream may still receive transactional mail.

Postmark: List Templates

List the templates and layouts on this server. A server holds at most 100 templates.

Postmark: List Webhooks

List the webhooks configured on this server, with their URLs, verification status and which events each one is subscribed to.

Postmark: Search Inbound Messages

Search messages received on the inbound stream. Note the default status filter is processed, so blocked and failed inbound mail does not appear until you ask for it by status.

Postmark: Search Outbound Messages

Search sent messages by recipient, sender, tag, subject, status, stream or date window. Use extra_query to filter on a metadata value, for example metadata_order-id.

Action nodes 43

Postmark: Archive Message Stream

Archive a message stream. Sending on it stops immediately, and Postmark purges it permanently on the returned ExpectedPurgeDate. Unarchive it before that date if this was a mistake; after the purge the stream and its data are gone.

Postmark: Bypass Inbound Block

Release a blocked inbound message so Postmark delivers it to your inbound webhook after all. Only works on messages whose status is blocked.

Postmark: Create Domain

Add a sending domain and get back the DNS records to publish. A custom Return-Path must be a subdomain of the From domain. Needs the account token.

Postmark: Create Inbound Rule

Block an address or a whole domain from delivering inbound mail. Blocking a domain silently stops every sender on it, so check the value before saving.

Postmark: Create Message Stream

Create a message stream. The type is fixed at creation and cannot be changed later, and Broadcast streams require unsubscribe management, so choose deliberately rather than reusing a transactional stream for campaigns.

Postmark: Create Sender Signature

Add a From address. Postmark emails that address a confirmation link and the signature cannot send until someone clicks it, so the mailbox has to be one a person actually reads. Needs the account token.

Postmark: Create Server

Create a server, which is Postmark's isolation boundary: its own API token, its own streams, its own statistics and its own sender configuration. Needs the account token.

Postmark: Create Template

Create a template or a layout on this server. Give it an alias so sends can reference it by a stable name instead of a numeric ID.

Postmark: Create Webhook

Create a webhook for delivery, bounce, open, click, spam complaint or subscription change events. Postmark verifies the endpoint before saving unless you turn that off, and it does not sign payloads, so protect the endpoint with HTTP basic auth or a secret in the URL.

Postmark: Delete Domain

Delete a sending domain. Every sender signature on it stops being able to send, and the domain must be added and verified again from scratch to recover. Needs the account token.

Postmark: Delete Inbound Rule

Remove an inbound block, so mail from that address or domain is accepted again.

Postmark: Delete Sender Signature

Delete a sender signature. Anything still sending from that address starts failing immediately, and re-adding it means going through email confirmation again. Needs the account token.

Postmark: Delete Server

Delete a server and everything scoped to it: its API tokens, message streams, templates, webhooks and message history. This cannot be undone, and Postmark does not enable it on every account. Needs the account token.

Postmark: Delete Template

Delete a template. Any automation still sending with this alias or ID starts failing immediately, and Postmark documents no way to restore it.

Postmark: Delete Webhook

Delete a webhook. Events stop being delivered immediately and are not replayed once a new webhook is created, so anything that fires in the gap is lost.

Postmark: Push Templates Between Servers

Copy templates from one server to another, matching on alias. Run it first with PerformChanges false to see exactly which templates would be created or overwritten. Needs the account token.

Postmark: Reactivate Bounced Address

Reactivate an address Postmark deactivated after a bounce, so sending to it resumes. Only do this when you know the underlying mailbox problem is fixed; repeatedly reactivating a dead address damages your sending reputation.

Postmark: Remove Suppressions

Remove addresses from a stream's suppression list so sending to them resumes. Spam complaint suppressions cannot be removed, by design, and those entries come back with a Failed status.

Postmark: Request Data Removal

Erase a recipient's data from the Postmark account, for GDPR and CCPA subject requests. This is irreversible: once the request completes, that recipient's message history, event data and content are permanently gone and Postmark offers no way to recover any of it. Access to this endpoint is granted by Postmark support on request. Needs the account token.

Postmark: Request New Sender DKIM

Request a new DKIM key for a sender signature. Postmark has deprecated this endpoint in favour of rotating DKIM on the domain, and returns an error if a renewal is already in progress. Needs the account token.

Postmark: Resend Sender Confirmation

Send the confirmation email for a pending sender signature again. Needs the account token.

Postmark: Retry Inbound Message

Requeue an inbound message whose webhook delivery failed. Fix the endpoint first, otherwise it fails again.

Postmark: Rotate Domain DKIM Key

Generate a new DKIM key pair for the domain. The new record appears as pending and must be published in DNS and verified before the old key can be safely removed; removing the old record early breaks signing. Needs the account token.

Postmark: Send Bulk Broadcast

Queue a bulk broadcast to many recipients from one shared body or template. Postmark keeps broadcast and transactional traffic apart on purpose, so this node requires a Broadcast message stream and refuses to run against a transactional one. The send is accepted and processed asynchronously, so check the bulk request status for the outcome.

Postmark: Send Email

Send one email with Postmark's own request shape, exposing every provider field: attachments, inline images, custom headers, metadata and per-message tracking. This is the raw passthrough send. For campaign sending use Postmark: Send Campaign Email, which keeps the same parameters across every email provider.

Postmark: Send Email Batch

Send up to 500 individually addressed emails in one call. Each message is validated and delivered independently, so the response is a per-message list and a partial failure is normal: check every entry's ErrorCode, not just the HTTP status.

Postmark: Send Email With Template

Send one email rendered from a stored template, supplying the values its placeholders need. Provide TemplateAlias or TemplateId, never both.

Postmark: Send Template Batch

Send up to 500 template-rendered emails in one call, each with its own template and its own model. Results come back per message, so a partial failure is normal and every entry's ErrorCode needs checking.

Postmark: Suppress Addresses

Add addresses to one stream's suppression list so Postmark stops sending to them. Results come back per address, so check each Status rather than assuming a 200 means all of them took.

Postmark: Unarchive Message Stream

Restore an archived message stream, provided its purge date has not passed.

Postmark: Update Domain

Change a domain's custom Return-Path. It must be a subdomain of the From domain and needs its own CNAME record. Needs the account token.

Postmark: Update Message Stream

Rename a stream, change its description or change how unsubscribes are handled. The stream type itself is not editable.

Postmark: Update Sender Signature

Update a signature's display name, Reply-To or Return-Path. The From address itself cannot be changed; create a new signature instead. Needs the account token.

Postmark: Update Server

Update the server this credential belongs to. Switching DeliveryType to Sandbox makes the server accept mail and silently discard it, which is useful for testing and damaging if left on by mistake.

Postmark: Update Server By ID

Update any server on the account by ID. Needs the account token. Switching DeliveryType to Sandbox makes that server accept mail and discard it.

Postmark: Update Template

Update a template's name, subject, bodies, alias or layout. The change applies to every future send that references it.

Postmark: Update Webhook

Change a webhook's URL, authentication, headers or which events it receives.

Postmark: Validate Template

Render template content against a test model without saving or sending anything, and get back the rendered output plus any syntax errors with line numbers.

Postmark: Verify Domain DKIM

Ask Postmark to re-check the domain's DKIM DNS record and update its verification status. Run this after publishing the record. Needs the account token.

Postmark: Verify Domain Return-Path

Ask Postmark to re-check the domain's custom Return-Path CNAME and update its verification status. Needs the account token.

Postmark: Verify Domain SPF

Re-check the domain's SPF record. Postmark has deprecated this endpoint because SPF is no longer required for its own sending; prefer DKIM and Return-Path verification. Needs the account token.

Postmark: Verify Sender Signature SPF

Re-check the SPF record behind a sender signature. Postmark has deprecated this endpoint; prefer DKIM and Return-Path verification on the domain. Needs the account token.

Postmark: Verify Webhook

Make Postmark post a test payload for each subscribed trigger and report the status code your endpoint returned for each one.

Connecting Postmark takes one credential

Both tokens are copied out of Postmark, and the reason people get stuck is that both pages are called API Tokens. The SERVER token is on a server: open Postmark, choose the server you are integrating with, and open its API Tokens tab. That is the token that covers sending, message search, statistics, bounces, templates, message streams, suppressions, webhooks and inbound rules, and it is what the connection test uses. The ACCOUNT token is on the account rather than on any server: open your Postmark account settings and its own API Tokens page. Postmark's API overview describes the server token as available to account owners, admins and server admins, and the account token as available to account owners and admins, while its account-level endpoint pages describe those endpoints as accessible by the account owner, so if you have no Account API Tokens page to open, somebody with owner access has to fetch that token for you. Leave the account token blank if you are not administering servers, domains or sender signatures. The 62 server-level nodes never look at it, and the 25 that need it refuse before making a request and tell you which token is missing and which page it is on, rather than sending the wrong header and collecting a 401 that reads like a bad token.

1

Decide which Postmark server the connection represents. A server is Postmark's isolation boundary: its own token, its own streams, its own templates and its own statistics. A connection is scoped to one, so a staging server and a production server are two connections rather than one with a switch.

2

Copy the server API token from that server's API Tokens tab in Postmark.

3

Note the Broadcast message stream this connection should use for campaign mail. The connection test checks it, and it must be a stream whose type Postmark reports as Broadcasts, not the transactional one.

4

Copy the account API token as well if you intend to use the account-level nodes for servers, domains, sender signatures, template push or data removal. It is on your Postmark account's own API Tokens page, which is a different page from the server's. Skip this if you do not need those 25 nodes.

5

In Mattering Automations, open Connectors, click Add Connector, choose Postmark, and create a new credential. Paste the server token, the broadcast stream id, and the account token if you have one. Tokens are stored encrypted and never appear in workflow logs.

6

Click Test Connection. Mattering reads the server with the server token and then reads the message stream you named and confirms Postmark reports its type as Broadcasts, so a pass proves the token and the stream together rather than the token alone.

7

All 87 Postmark nodes appear in the canvas palette. Postmark: Get Server is the cheapest thing to run first, since it takes no parameters and spends no email allowance, and Postmark: List Message Streams is the second, because it shows you every stream on the server with its type.

Common uses for the Postmark MCP

A morning deliverability report that reaches somebody

Postmark: Get Sending Overview gives sent, bounced, spam complaints, tracked mail, opens and clicks with the bounce and complaint rates for a window, and Postmark: Get Spam Complaint Counts is the daily line that matters most, because a rising complaint rate is the earliest reliable warning that a sending reputation is going wrong. Postmark: Get Delivery Stats adds the lifetime bounce totals and how many addresses Postmark has currently deactivated. An AI step can turn the numbers into two paragraphs of plain English, and a Productivity task or a Slack message puts them where someone reads them. Build the window in Eastern time: Postmark evaluates these date filters in United States Eastern rather than UTC, and rates are percentages of sent, so they must never be added across days.

Bounce triage that keeps the list clean without guessing

Postmark: List Bounces filters by bounce type, address, tag, message, stream and date window, and the Inactive flag on each row is the operational one: it tells you whether Postmark has already stopped sending to that address. Postmark: Get Bounce reads the receiving server's own explanation, and Postmark: Get Bounce Source returns the raw SMTP conversation for the cases where the summary is not enough, though Postmark keeps those dumps for 30 days only. Where a mailbox problem is genuinely fixed, Postmark: Reactivate Bounced Address resumes sending, and the node says plainly that reactivating a dead address repeatedly damages your reputation. Feed the hard bounces back into whatever holds your list, so the same address is not sent to again from the other side.

Broadcast sending that cannot land on the transactional stream

Postmark: Send Bulk Broadcast queues one shared body or template to many recipients and returns a request id rather than a result, because Postmark processes it in the background. Postmark: Get Bulk Request then reports status, percentage complete, released count and failed count, which is what a workflow should wait on rather than assuming acceptance means delivery. The stream field on this node offers Broadcast streams only, refuses the reserved transactional id outright, and reads the chosen stream back to confirm Postmark reports its type as Broadcasts before anything is sent. That is the whole point: bulk mail on a transactional stream is the failure Postmark's stream model exists to prevent, and it stays invisible until receipts and password resets start landing in spam.

Domain and sender setup driven from a checklist rather than a memory

Onboarding a new sending domain is a sequence people get wrong once a year and never remember. Postmark: Create Domain returns the exact DNS records to publish, Postmark: Verify Domain DKIM and Postmark: Verify Domain Return-Path ask Postmark to re-check them once they are live, and Postmark: List Domains reports the verification state of every domain on the account in one call, which makes a weekly audit trivial. Postmark: Create Sender Signature adds a From address and Postmark: Resend Sender Confirmation chases the confirmation email that somebody has to click before it can send. Postmark: Rotate Domain DKIM Key handles key rotation, and its description carries the rule that matters: publish and verify the new record before removing the old one, because removing it early breaks signing. All of these need the account token.

Answering the question of what actually happened to one email

A customer says they never got the receipt. Postmark: Search Outbound Messages finds it by recipient, tag, subject, status, stream or date, and supports filtering on a metadata value, so an order id attached at send time is enough to find the message without knowing anything else. Postmark: Get Outbound Message returns the whole event timeline of deliveries, transient failures, opens, clicks, bounces and subscription changes, and Postmark: Get Outbound Message Source returns the raw MIME for the cases where a receiving mail server disputes what was delivered. An assistant with these three read nodes and nothing else can answer most support escalations without anyone opening the Postmark interface, and it cannot send, suppress or delete anything while doing it.

Inbound mail that gets processed rather than silently blocked

Postmark: Search Inbound Messages reads what arrived on the inbound stream, and its status filter has a trap the node names: Postmark defaults it to processed, so blocked and failed mail is invisible until you ask for it by status. That is exactly the sweep worth automating. Postmark: Get Inbound Message shows the full message and, where it was blocked, which rule blocked it. Postmark: Bypass Inbound Block releases a legitimate message that a rule caught, and Postmark: Retry Inbound Message requeues one whose webhook delivery failed, once the endpoint is fixed. Postmark: List Inbound Rules and Postmark: Create Inbound Rule manage the blocking rules themselves, with the caution that blocking a whole domain stops every sender on it at once.

Example: A weekly sending health review that catches problems before recipients do

Built on the Automations canvas in a few minutes, or described in plain language to Matter and built for you.

1

Trigger: a weekly schedule fires on Monday morning. This connector has no trigger nodes, so Postmark workflows run on a schedule and read, rather than waiting for a push.

2

Postmark: Get Sending Overview reads the past seven days: sent, bounced, spam complaints, tracked mail, opens, clicks, and the bounce and complaint rates. The window is built in Eastern time, because that is the zone Postmark evaluates these filters in.

3

Postmark: Get Spam Complaint Counts and Postmark: Get Bounce Counts return the same period day by day, so a single bad send is distinguishable from a slow drift, which the overview alone cannot show.

4

Postmark: List Bounces pulls the hard bounces from the window, and Postmark: List Suppressions reads the suppression list on the broadcast stream so the two can be compared.

5

A condition checks whether the complaint rate or the bounce rate has crossed the threshold your team agreed on. Below it, the workflow writes the figures to a Mattering data table and ends.

6

Above it, an AI step drafts the summary and the likely cause from the daily shape and the bounce reasons, and a Productivity task is created for a person to read. Nothing is suppressed or deactivated automatically, because the response to a reputation problem is a decision rather than an action.

7

On approval, Postmark: Suppress Addresses adds the agreed addresses to the stream's suppression list. Note that spam complaint suppressions cannot be removed afterwards, by Postmark's design, which is another reason this step waits for a person.

Postmark MCP questions, answered

What can this connector actually do in Postmark?

87 nodes: 44 that read and 43 that act, across the fifteen sections of Postmark's REST API. Sending covers one message, a batch of up to 500, a template render, a template batch and an asynchronous bulk broadcast with its request status. Messages covers outbound and inbound search, full detail with the event timeline, raw MIME, and per-message opens and clicks. Statistics covers a sending overview plus eleven daily breakdowns. Operations covers bounces and reactivation, per-stream suppressions, message streams including create, edit, archive and unarchive, webhooks including verification and delivery statistics, and inbound blocking rules. Administration covers servers, domains with DKIM and Return-Path verification and DKIM rotation, sender signatures, template push between servers, and data removal. What is not here: Postmark's webhooks are managed by these nodes rather than being Mattering trigger nodes, so a Postmark workflow runs on a schedule and reads. The connector also unlocks the shared campaign send node, postmark_send_email, which keeps one uniform contract across email providers for Sales campaign sending and is separate from the provider-shaped Postmark: Send Email node.

Does this work with Claude, ChatGPT, or other AI assistants?

Yes. Mattering exposes these nodes over MCP, an open protocol for connecting AI assistants to tools, so any MCP-capable assistant, including Claude and ChatGPT, can connect to your workspace and work with Postmark through them. Matter, the assistant built into the platform, uses the same nodes natively. The same nodes are also steps on the automation canvas, so a workflow an assistant helps you design can then run on a schedule with nobody present. The sensible split for email is to hand an assistant the read nodes, which is where most of the value is anyway, and keep sending, suppression and the eight destructive nodes behind a condition or an approval step.

Postmark publishes its own MCP server. Why use this one?

It does, and it is worth knowing about: github.com/ActiveCampaign/postmark-mcp, maintained under Postmark's parent company and linked from Postmark's developer page. It exposes 24 tools covering sending, templates, message search, bounces, suppressions, webhooks, delivery stats and server info, it runs locally next to your assistant, and for an interactive assistant that only needs those things it is a good fit. Three differences decide the other cases. Coverage: that server is configured with a server token, a sender address and a stream, so the 25 account-level nodes here, which are servers, domains, sender signatures, template push and data removal, plus the daily statistics breakdowns, message stream management, bulk broadcast, inbound handling and inbound rules, are outside it. Autonomy: a node is a step in a workflow, so it runs on a schedule with nobody present and chains with every other connector in the same run, which a local assistant tool cannot do. Custody: the credential lives once in a shared workspace rather than in each person's config, and destructive operations are flagged as destructive rather than being one more callable tool.

Why does Postmark need two tokens, and what breaks if I only add one?

Postmark authenticates server-level and account-level requests with two different headers, X-Postmark-Server-Token and X-Postmark-Account-Token, and they are not interchangeable. The server token covers sending, messages, statistics, bounces, templates, message streams, suppressions, webhooks, inbound rules and the server itself, which is 62 nodes. The account token covers servers, domains, sender signatures, pushing templates between servers and data removal, which is 25. If you add only the server token, those 62 nodes work normally and the other 25 refuse before making any request, with a message that names the token, says which page it is on, and lists which nodes do and do not need it. Nothing reaches Postmark, so there is no 401 to misread. The confusing part in Postmark itself is that both pages are called API Tokens: the server one is on the server, the account one is on the account. Postmark's overview describes the account token as available to account owners and admins, so an account with restricted access may need an owner to fetch it.

Can I send a newsletter through this, and what stops me getting it wrong?

Yes, through Postmark: Send Bulk Broadcast, and the thing that stops you getting it wrong is the stream. Postmark separates Transactional, Broadcast and Inbound streams so that bulk mail cannot damage the deliverability of receipts and password resets, and Postmark's own send endpoint defaults to the reserved transactional stream, named outbound, when no stream is given. So on the bulk node the stream is required with no fallback, the picker offers Broadcast streams only, the reserved outbound id is refused by name before any request is made, and the stream you chose is read back and rejected unless Postmark reports its type as Broadcasts. There is no override flag, deliberately. Bulk sending is also asynchronous: the node returns a request id, and Postmark: Get Bulk Request reports status, percentage complete, released count and failed count, which is what a workflow should actually wait on.

What does Postmark cost, and can I test this for free?

Postmark starts new accounts on a free developer plan offering 100 emails a month to test an integration with, which Postmark states never expires or runs out, and which includes the REST API, SMTP and Sandbox Mode. That is enough to build and verify a workflow properly. For exercising the send nodes without spending the allowance, set a server's delivery type to Sandbox: it accepts mail and discards it, so nothing reaches a real inbox. Postmark: Get Server, Postmark: List Message Streams and every statistics and message read node cost no email at all. Current paid volumes and prices are on Postmark's own pricing page and we are not going to restate them here, because they change and a stale number on this page would be worse than no number.

What are Postmark's rate limits?

Postmark documents a 429 Rate Limit Exceeded response for making requests at a rate that exceeds acceptable use of the API, and does not publish a specific requests-per-minute number, so we are not going to invent one. What is documented and worth designing around: an oversized request is refused with a 413, which Postmark's overview documents as a 10 MB limit while its batch and bulk endpoints document a 50 MB payload allowance; the batch endpoints accept up to 500 messages per call; and list endpoints cap count at 500 with count plus offset unable to exceed 10,000, which is a hard ceiling on how deep a paged walk can go rather than a speed limit. This connector does not auto-paginate, so each node is one request and an unbounded number of calls never hides behind a single canvas step.

How far back can I search messages and statistics?

As far as your retention period, which Postmark documents as 45 days by default and configurable from 7 to 365 days. Message data expires at the end of it, so a search reaching past that window returns nothing rather than an error, which is a normal answer that looks like a broken node. Two other retention details are worth knowing: Postmark does not store bounce dumps older than 30 days, so Postmark: Get Bounce Source returns nothing for older bounces, and Postmark: Get Webhook Statistics reports on Postmark's own fixed trailing 24 hours, which is not selectable. And build every window in United States Eastern time rather than UTC, because that is the zone Postmark evaluates fromdate and todate in on statistics, messages, bounces and suppressions.

Which nodes can cause damage, and how are they marked?

Eight are classed destructive and each says what it costs in its own description. Postmark: Request Data Removal is the most serious: it erases a recipient's message history, event data and content for a GDPR or CCPA subject request, Postmark documents no way to recover any of it, and Postmark makes the endpoint available by request through support rather than enabling it by default. Postmark: Delete Server removes a server's tokens, streams, templates, webhooks and message history. Postmark: Delete Domain stops every sender signature on that domain, and Postmark: Delete Sender Signature makes anything still sending from that address fail immediately. Postmark: Archive Message Stream stops sending at once and Postmark deletes archived streams 45 days after archiving, restorable with Postmark: Unarchive Message Stream only until then. Postmark: Delete Template breaks any send still referencing that alias or id. Postmark: Delete Webhook stops events immediately with no replay, so anything firing in the gap is lost. Postmark: Delete Inbound Rule is the mildest, since it only unblocks a sender.

87 nodes waiting on the canvas

Put Postmark to work inside your workspace

Connect once, then use every node in visual automations and AI agent workflows across the Mattering platform.

All product names, logos, and brands are property of their respective owners and are used for identification purposes only. Use of these names, logos, and brands does not imply endorsement or affiliation. Postmark is a product of its respective owner; Mattering is an independent platform.

Start Free Explore our Products