For a while now, MeshMonitor's automations have been a fixed menu: Auto Acknowledge, Auto Traceroute, Auto Ping, Auto Responder, Auto Announce. Each one does a single, well-defined job. They're great — but if you wanted something they didn't already do, you were out of luck.

Not anymore. MeshMonitor 4.12.0 introduces the Automation Engine — a generic, visual "when this happens, do that" builder, in the spirit of Home Assistant, Node-RED, and IFTTT. It lives on its own top-level Automations tab, and it lets you wire up your own behavior instead of relying on the hardcoded ones. The legacy automations remain exactly as they were; the engine is the build-it-yourself alternative.

The shape of an automation

Every automation is a small graph you build in a guided, linear editor:

WHEN  →  RULE (IF … THEN …)  →  optional FINALLY

• WHEN — exactly one trigger that starts things off.
• RULE — one or more conditions that decide whether to act, each routing to its own actions. Conditions are routers with a true and false path, so you get If / ElseIf / Else logic.
• FINALLY — an optional combine step that runs based on how the rules turned out: ANY, ALL, NONE, or ALWAYS.

Automations run globally — they evaluate events from every connected source at once, just like Map Analysis — with an optional Source filter condition when you want to scope one down. They're gated by a dedicated automations permission, carry a per-automation cooldown and loop guard, log every run for debugging, and export/import as JSON (imports land disabled for review).

Triggers, conditions, actions

Triggers — a message arrives (with channel-name matching and substring/regex filters), a node is discovered or updated, telemetry crosses a threshold, a cron schedule fires (#3726), a system event happens (start, source online/offline, upgrade available), or a node enters/leaves a geofence drawn right on the map (#3721).

Conditions — number and text comparisons over the event, the hydrated node record (battery, hops, role, position, age…), or its latest telemetry; a source filter; distance from a point; a variable check; and a time-of-day window. A missing field never throws — it just evaluates false.

Actions — send a tapback, send a message (with a "Send via sources" multi-select and unified "On channels" picker that badges MeshCore vs Meshtastic), manage a node, fire an Apprise notification, or run a script from your data directory with MM_* environment variables and a JSON-typed result you can reference later (#3746).

Variables and tokens

You can define your own variables — constants or runtime flags/counters — in a dedicated management area, with global/source/node scopes and nested access into JSON values. And everywhere you can type text, you can drop in substitution tokens that resolve at run time:

Node {{ trigger.fromId }} just reported {{ trigger.value }} at {{ NOW }}.
Last seen counter: {{ var.seenCount }}

The builder validates tokens as you type — recognized tokens turn blue, typos get a red wavy underline (#3727, #3739) — and a Substitutions drawer lists every token available for the current trigger.

Test before you ship

Every automation has a built-in Test panel: run it against a synthetic event — set SNR/RSSI/Via-MQTT, hops-away, node facts — with no mesh traffic, no notifications, and nothing saved. The panel shows the interpolated message, the chosen action, and, when nothing fires, explains which condition went false and what to change.

Try it

The Automation Engine is in the new Automations tab in 4.12.0. The full guide — triggers, conditions, actions, variables, tokens, and the test panel — is in the Automation Engine documentation.

This feature shipped across a series of PRs: #3653 (the engine), #3721, #3726, #3727, #3739, and #3746.