Automations
A trigger fires, conditions decide whether it matters, actions run. Automations replaced the six fixed rule types in 2.2; the same detections ship as templates you can open and change.
Trigger, conditions, actions
There are two kinds. A policy records a violation against the person it fired on. A notification only runs its actions. Any trigger, condition, group, or action can be switched off without deleting it.
Triggers
Eighteen of them across six groups. One automation can listen to several at once, and the variables you can put in a message are whatever all of them offer.
Conditions
Thirty-one fields, ten operators, and groups that match all or any of their rows. Groups themselves are combined with AND.
Actions
Four things an automation can do, run in the order you list them, plus an if step that branches on its own conditions.
18 triggers across 6 groups
Sessions
- A stream starts
- A stream is first seen
- A stream stops
- A stream is paused
- A stream has been paused for N minutes, this time or in total
- Transcoding starts or stops
Accounts
- An account has been inactive for N days
- An account plays from a new device
- An account's trust score changes
Library
- Media is added
- Media is upgraded
Servers
- A server goes down
- A server comes back up
Updates
- A plugin update is available
- A server update is available
- A Tracearr update is available
Notifications
- A newsletter is sent
- A newsletter send fails
31 condition fields across 7 categories
Counting conditions read the whole person, not one account. Concurrent streams, distance between sessions, travel speed, and unique IPs or devices in a window all aggregate across every account a merged identity holds.
Session behaviour
- Concurrent Streams
- Distance Between Sessions
- Travel Speed
- Unique IPs in Window
- Unique Devices in Window
- Days Inactive
- Current Pause Duration
- Total Pause Duration
Stream quality
- Source Resolution
- Output Resolution
- Is Transcoding
- Transcode Downgrade
- Source Bitrate
Account
- User
- Trust Score
- Account Age
Device and client
- Device Type
- Client/Player
- Platform/OS
Network and location
- Local Network
- Country
- IP Range
Scope
- Server
- Media Type
Media
- Item Type
- Library
- Resolution
- Dynamic Range
- Video Codec
- Audio Channels
- File Size
Actions
- Send notification. Send to one or more destinations, with your own title and body and any of the trigger's variables written into them.
- Trust score. Adjust a score by an amount, set it outright, or reset it. Needs an account in context.
- Kill stream. Terminate the session after a delay of up to five minutes, with a message the player shows. The condition is checked again before the kill, so a stream that sorted itself out is left alone.
- Message client. Send a message to the player without stopping it. Jellyfin and Emby only, since Plex will only carry a message attached to a kill.
Send, trust, and stop stream each carry their own cooldown, and the automation has one of its own on top.
Nine destinations
Seven you configure, plus mobile push and browser toasts, which are always there. You can add as many of each as you like and point one automation at several. Credentials are encrypted at rest and never read back out.
Start from a template
Twenty-two ship with the app. A template asks for the values it needs, such as a server, an account, a destination, or a limit, and creates a normal automation you can open and edit.
Notifications
Stream started, New media added, New device
Policies
Too many streams at once, Impossible travel, Blocked countries
Server health
Server down, Media server update, Tracearr update
Housekeeping
Stop paused streams
Share codes
Export any automation and you get one line of text starting with tracearr1. that somebody else can paste into their install. Server ids, account ids, destinations, and IP ranges are lifted out and turned into questions the importer answers, and an export carrying a raw id is refused rather than shipped. Nothing phones home: the code is the whole payload, and an import arrives switched off so you can read it first.
Live check
While you are editing, run the draft against the streams playing right now. You get each session it looked at, which conditions passed with the numbers behind them, and which actions would have run. No run is recorded and nothing executes. Cooldowns and the duplicate gate are not simulated, so a sample that says it would run can still be held back for real.
Runs
Each run records the trigger payload, every condition with the value it saw, the actions that ran, and the session behind it: title, player, device, IP, and city. Completed policy runs are violations, so the violations list and an automation's activity tab show the same rows.
- Editing an automation writes a new version, and a past run still shows the definition that actually ran
- Near misses are kept too, with the reason: a cooldown, a repeated edge, the duplicate gate, or a trigger filter
- Retention is per automation, defaulting to a year for policies and thirty days for notifications