MQTT

· #96 most-used

The pub/sub backbone for every connected device

DeveloperAutomationCommunicationAnalyticsSecurity

MQTT is an OASIS-standard publish-subscribe protocol built for IoT — tiny payloads, minimal bandwidth, and reliable delivery even over lossy 4G links. Connect it to Actionist and your agents can publish commands to devices, subscribe to live sensor feeds, read retained state without waiting for new events, and react to broker anomalies the moment they appear. From a single smart thermostat to a 50,000-device industrial fleet, the agent is always listening — and always ready to act.

Average time saved
10 hours
per person · per month
1 workdays back

Eliminates manual work. Eliminates manual broker log-trawling, spreadsheet-based device health checks, and hand-crafted publish scripts that ops engineers run every morning.

Schedule

What your MQTT agent runs on autopilot

A week of scheduled jobs your Actionist agent will execute on your behalf.

28Scheduled jobs
7Agents at work
24/7Always on
Mon
Tue
Wed
Thu
Fri
7am
8am
9am
10am
11am
12pm
1pm
2pm
3pm
4pm
5pm
6pm
Agents
Multi-app workflows

MQTT × every other app you use

End-to-end automations that span multiple apps — each one a real business outcome.

6Workflows
9Apps spanned
~41 hrsSaved / week
6Personas served
customer-success★ FeaturedSaves 45m saved · runs ~15× /week

IoT alert to resolved ticket in 60 seconds

When a customer's device publishes a critical error payload to the MQTT broker, your agent reads the retained error state, opens a support ticket in Gmail, publishes an acknowledgement back to the device topic, posts the ticket link in Slack, and books a follow-up call on Google Calendar — all before a human has finished reading the alert email. Customer confidence stays high; response SLAs stay green.

Trigger: When a device publishes a message to the `devices/<id>/error` MQTT topic
Step 1 trigger
Gmail
Receive error notification email from device fleet
Step 2 read
Mqtt
Get retained message from device error topic
Step 3 write
Mqtt
Publish acknowledgement to device command topic
Step 4 write
Slack
Post ticket details to #support-ops channel
Step 5 write
Google Calendar
Create follow-up call with customer within 24 hours
Zero missed device alerts
ROI

Savings

What your team gets back — two angles: what you stop doing manually, and what that's worth.

Without Actionist

What you do manually today

With Actionist

What your agent runs for you

  • Sales
    18 min / week
    Trial usage tracking

    Sales reps manually check broker dashboards to see whether trial devices are active, usually once a day at best.

    Sales Agent
    0 min
    Agent flags high-intent device signals

    Agent reads broker stats for every trial client ID and posts engagement spikes to #iot-trials before the rep's morning coffee.

  • Marketing
    13 min / week
    Campaign reach reports

    Marketers export device-count CSVs from the broker admin panel and paste them into campaign decks by hand.

    Marketing Agent
    0 min
    Agent publishes and logs fleet broadcasts

    Agent publishes announcement payloads to the fleet and writes delivery counts to the campaign tracker in real time.

  • Customer Support
    18 min / week
    Reactive error polling

    Support engineers SSH into broker hosts and grep logs to find the last error payload a device published — often 20 minutes after the customer calls.

    Customer Support Agent
    0 min
    Agent reads retained error state instantly

    Agent fetches the retained error topic for any device in seconds and opens a pre-populated ticket before the call ends.

  • Human Resources
    7 min / week
    On-call roster vs. broker alerts

    HR maintains a separate spreadsheet to cross-reference which engineer is on call when the broker fires an alert — manual lookup every incident.

    Human Resources Agent
    0 min
    Agent routes alerts to the on-call engineer

    Agent reads the on-call schedule and publishes a direct-channel MQTT command to the right engineer's notification device.

  • Finance
    13 min / week
    Metered usage reconciliation

    Finance pulls message-count CSVs from the broker monthly and manually maps them to customer invoices — error-prone and hours-long.

    Finance Agent
    0 min
    Agent reads broker stats at billing close

    Agent fetches per-client message volumes from $SYS topics and populates invoice line items in HubSpot automatically.

  • Operations
    25 min / week
    Fleet config push via terminal

    Ops engineers SSH into each device group and run mosquitto_pub scripts to distribute config updates — a 25-minute ritual per release.

    Operations Agent
    0 min
    Agent publishes retained config to fleet

    Agent reads the config sheet, publishes retained payloads to each device topic, and logs every delivery to the audit trail.

  • Legal
    6 min / week
    Data-residency audit

    Legal manually documents which MQTT topics carry PII by interviewing engineers and reading broker logs — slow and incomplete.

    Legal Agent
    0 min
    Agent maps topic-to-data-type inventory

    Agent subscribes to topic patterns, classifies payload schemas, and writes the data-residency map to a Notion compliance page.

+ 100s of other MQTT automations
Average monthly
10 hrs / person / month
Average monthly
10 hrs / person / month
Calculator

Calculate what your team saves

Team size
10 people
Hourly rate
$20 / hr
Hours saved / week
25
Hours saved / year
1,250
Annual ROI
$25,000

Based on MQTT's typical team usage — the visible tasks plus a few other automations the agent runs: ~2.5 hrs / person / week of admin work automated.

Connect

How to plug MQTT into Actionist

Pick the connection method that suits your environment.

The fastest path. Install Actionist's MQTT MCP server in one click and the agent publishes, subscribes, and reads retained messages through a permissioned broker session — no raw credential strings to manage.

1
Open the Apps tab

Find MQTT in the Apps library and click Connect. MCP is selected by default.

2
Enter broker details

Provide the broker host, port, and protocol (mqtt://, mqtts://, ws://, or wss://). Actionist stores credentials encrypted; you never paste them into a workflow node.

3
Test the connection

Actionist runs a read-only call to verify the handshake. You're ready.

Actions

16 actions your agent can call

Read and write operations available to your Actionist agent.

Triggers

7 events your agent can react to

Events your agent watches for, and the actions it kicks off in response.

Skills

Skills that pair with MQTT

Reusable agent skills that work well alongside this app.

Uptime Kuma

Check monitor status, pause/resume uptime checks, and view heartbeat history directly from a workflow — pairs naturally with MQTT broker health monitoring.

MCP servers

MCP servers that work with MQTT

Connect Actionist to MCP servers built for or around this app.

DDC/CI Control Bridge
Official

Control monitor hardware (brightness, contrast, input) via DDC/CI — uses MQTT as the transport layer for remote display commands.

getmockd/mockd

Mock MQTT brokers alongside HTTP, GraphQL, and WebSocket in one server — ideal for testing MQTT workflows without a live broker.

mockd
Official

Official multi-protocol mock server supporting MQTT, WebSocket, HTTP, GraphQL, gRPC, SSE, and SOAP for integration testing.

FAQs

Questions about MQTT + Actionist

How do I connect Actionist to my MQTT broker?
In the Apps tab, find MQTT and click Connect. Choose MCP for the fastest path — Actionist handles the broker session for you. Alternatively, select Broker credentials and enter your hostname, port, and username/password. TLS brokers: enable SSL and paste your CA certificate. The agent runs a test subscribe on connection to confirm the handshake before any workflow runs.
What credentials does the agent need to publish and subscribe?
For password-authenticated brokers, the agent needs a dedicated MQTT username and password with publish and subscribe permissions on the relevant topic patterns. For TLS brokers (port 8883), also provide the CA certificate — and optionally a client cert/key pair if your broker requires mutual TLS. Avoid using the broker's admin account; create a scoped client user instead.
Can the agent combine MQTT with other apps like Slack or Google Sheets?
Yes — MQTT is load-bearing in every Actionist workflow alongside apps like Slack, Google Sheets, HubSpot, GitHub, and Notion. A typical pattern: the agent subscribes to a sensor topic, reads the payload, writes a row to Google Sheets, and posts a summary to Slack — all in a single workflow with no custom code. Any action in the MQTT catalog can be chained with any other connected app.
What are the most common MQTT automation use cases in Actionist?
The most-used patterns are: IoT device error → support ticket (agent reads retained error state, opens ticket, publishes ack); scheduled fleet health digest (agent reads heartbeats, writes Google Sheets, flags missing sensors); config rollout from a spreadsheet (agent publishes retained config payloads to each device topic); and usage-based billing (agent reads $SYS broker stats at month-end and populates invoice line items).
How do I avoid trigger loops when the agent both reads and publishes on the same broker?
Publish to a dedicated agent-output topic that your workflow's trigger topic does not overlap — e.g. subscribe to `devices/+/error` but publish acks to `devices/+/ack`. Never subscribe to the same topic you publish to within one workflow. If you need the agent to react to its own output, use a separate workflow with a distinct trigger topic and add a `source: agent` field to payloads so the second workflow can filter non-agent messages out.
Which QoS level should I use for agent-published commands?
Use QoS 1 (at-least-once) for most agent commands — it guarantees delivery without the two-phase handshake overhead of QoS 2. Reserve QoS 2 (exactly-once) for safety-critical actuations like relay triggers or lock commands where a duplicate is dangerous. Use QoS 0 (fire-and-forget) only for high-frequency telemetry where a dropped message is acceptable and latency matters more than guarantee.
Does the agent support wildcard topic subscriptions?
Yes. Use `+` to match one level (`factory/line/+/temperature` matches any machine on any line) and `#` to match all subsequent levels (`devices/gateway-01/#` matches every sub-topic under that gateway). The agent receives all matching messages in a single subscription call and can route them by topic string inside the workflow. Avoid bare `#` subscriptions on high-volume brokers — message rate can exceed workflow throughput.
Can the agent read the last known value of a topic without waiting for a new message?
Yes — use the Get retained message action. It returns the last payload published with the retain flag set, along with the topic string, QoS, and timestamp. If no retained message exists on the topic, the action returns null rather than blocking. This is the right pattern for dashboard cold-starts, device-state lookups in support tickets, and config snapshot reads during CI jobs.