Notifications
Skynet tells people when something happens that they care about: a telescope started on an observation they are following, a download they asked for is ready, a device at their organization’s telescope faulted. This page explains which events produce notifications, who receives each one, and how they are delivered. To change your own settings, see Notification settings; to decide who on a telescope team hears about faults, see Operations notifications.
Three ways an event reaches a person
Section titled “Three ways an event reaches a person”Every notification starts as an event: a record Skynet writes when something changes state. Which people an event reaches depends on what kind of event it is, and the three routes never mix:
| Route | Used for | Who receives it |
|---|---|---|
| Observation watch | Observation events | Each person who turned on Notify for that observation. |
| Notification rule | Operational events (faults, connectivity, shutter, weather and safety) | The people the owning organization’s rules name: a person, a group, or everyone holding a role. |
| Addressed | Events about one person’s own request | That one person, and nobody else. |
Observation watches
Section titled “Observation watches”A watch is personal: you turn it on or off for yourself with the Notify bell in the observation’s header (see Follow an observation), or with Notify me when it’s complete on a project page. Anyone who can view an observation can watch it, and the page does not show who else is watching.
Skynet does not start a watch for you. Submitting an observation, owning it, or funding it does not subscribe you to it; press Notify on each observation you want to hear about. The same applies to observations an alert subscription creates.
Each time an observation event is delivered, Skynet checks again that the watcher can still view the observation. If the observation was made private or your access was removed, its notifications stop reaching you.
Notification rules
Section titled “Notification rules”Operational events are about telescopes and the hardware around them, so the organization that owns that hardware decides who hears about them. A notification rule names three things: where (the whole organization, one observatory, or one telescope), what (one or more operational families, below) and who (one member, a group, or everyone holding an organization role, resolved when each event is delivered). A rule never reaches across ownership: it only matches events on telescopes, observatories and devices its own organization owns.
Every organization starts with a default rule: device faults and instrument and telescope faults anywhere in the organization go to its owners. The default rule is an ordinary rule that managers can edit, pause or delete.
Rules are managed by organization owners and device managers. People who use a telescope through an access grant, but are not members of its organization, do not receive its operational notifications. Each person a rule reaches can mute it for themselves; nobody else can undo that mute. When several rules reach the same person about the same event, they are told once.
Setting rules up is covered in Operations notifications.
Addressed events
Section titled “Addressed events”Some events are about one person’s request rather than about the resource. A packaged download being ready is the example: it goes to the person who asked for it, even if others are watching the same observation, and it never appears in the observation’s Activity feed. A test sent from a notification rule is also addressed, one copy to each person the rule reaches.
Events and who they reach
Section titled “Events and who they reach”Each event belongs to a family. Families are what you choose channels for in your settings, and what notification rules name. The default channels apply until you change them.
| Event | Family | Who receives it | Default channels |
|---|---|---|---|
| A telescope started working on the observation | My observations | People watching the observation | In-app, email |
| The observation completed | My observations | People watching the observation | In-app, email |
| An alert’s refined position was applied to an alert-created observation | My observations | People watching that observation | In-app, email |
| An alert was retracted (the message says whether the observation was canceled or left active) | My observations | People watching that observation | In-app, email |
| Your packaged download is ready, or could not be packaged | My observations | The person who requested it | In-app, email |
| A device raised or cleared a fault | Device faults | People named by the owner’s rules | In-app, email, SMS |
| An instrument or telescope raised or cleared a fault (for example, an imaging chain producing unusable frames) | Instrument & telescope faults | People named by the owner’s rules | In-app, email, SMS |
| A device went offline or came back online | Device online / offline | People named by the owner’s rules | In-app, email |
| An enclosure shutter finished opening or closing | Enclosure shutter | People named by the owner’s rules | In-app, email |
| A weather or safety constraint became UNSAFE, SAFE, UNKNOWN or FAULT | Weather & safety | People named by the owner’s rules | In-app, email, SMS |
| A manager sent a test of a rule | The rule’s families | Everyone the rule reaches who has not muted it | Every channel the person gets for any of the rule’s families |
A few details about when these fire:
- Telescope started work fires when a telescope begins a task for the observation and the task it ran before was for a different observation, so a long run of tasks on one telescope announces itself once. It fires at most once every 10 minutes per telescope and observation.
- Enclosure shutter fires only on reaching fully open or fully closed, not for opening, closing or partly open states.
- Repeated changes are collapsed. The same fault (device, instrument or telescope) raised again within 5 minutes, and the same connectivity, shutter or weather change within one minute, produce one notification.
- Late events are dropped. If Skynet’s notification service is delayed and reaches an event more than an hour after it happened, that event is not sent, so a long outage does not end in a flood of stale messages.
Some events appear only in an observation’s Activity feed and never produce a notification: each individual task starting, finishing, or stopping before it finished. A busy observation can run hundreds of tasks, and these would bury its watchers.
Where a rule applies
Section titled “Where a rule applies”A rule’s scope contains everything inside it:
- Whole organization: every observatory and telescope the organization owns, including ones added later.
- One observatory: the observatory’s own devices (such as its weather sensors) and every telescope at it.
- One telescope: the telescope, its devices and instruments, and its enclosure. An event on a shared enclosure reaches the rules of every telescope inside it.
Weather and safety constraints that belong to the observatory rather than to one telescope are reached by observatory and whole-organization rules, not by telescope rules.
Delivery channels
Section titled “Delivery channels”Channels belong to each person, not to rules or watches: you decide, per family, how you hear about it.
| Channel | Where it arrives | Notes |
|---|---|---|
| In-app | The bell in the site header (an unread count, refreshed every minute) and the inbox at Dashboard → Notifications | Turning in-app off for a family keeps it out of the bell and the inbox; email and SMS for that family are unaffected. |
| Your account’s email address | The subject starts “Skynet —” and the message has a View on Skynet button that opens the related page. | |
| SMS | A phone number you have verified with a one-time code | Messages are cut to 160 characters. SMS is never sent to an unverified number, whatever your settings say. Reply STOP to stop all SMS at the carrier. |
How preferences resolve, for each family:
- A family you have never changed uses its default channels from the table above.
- A family you changed uses exactly the channels you ticked. Unticking every channel turns the family off.
- Muting a rule stops that rule reaching you; other rules that reach you are unaffected.
- The API also offers a pause on everything until a chosen time (
mutedUntilon your preferences); the website has no control for it. While it is set, nothing is delivered to you, and the events from that time are not sent later.
The website uses these endpoints; they are listed in the API reference under Notifications.
| Endpoint | Purpose |
|---|---|
GET /v1/notification-families |
The families, their labels and default channels, in display order. |
GET / PATCH /v1/me/notification-preferences |
Your per-family channels (eventFamilyChannels) and mutedUntil. Phone numbers are set only through /v1/me/phone/verify/start and /check. |
GET /v1/me/notifications |
Your in-app inbox, newest first (isRead filters it); mark read, mark all read, delete and clear read live beside it. |
GET /v1/me/subscriptions, POST /v1/entities/{slug}/subscriptions, PATCH / DELETE /v1/subscriptions/{uid} |
Your observation watches. You can only create a watch for yourself, on an observation you can view, for observation.* events. |
GET /v1/me/notification-rules, PUT / DELETE .../{rule_uid}/mute |
The rules that reach you, what you receive from each, and your mute. |
/v1/entities/{slug}/notification-rules, .../notification-coverage, .../{rule_uid}/test |
Managing an organization’s rules; needs the notifications.manage permission. |
Reference
Section titled “Reference”Event: one emitted event; drives notifications and the observation Activity feed.EventSubscription: an observation watch.NotificationRule,NotificationRuleMute: operational routing and a person’s opt-out from one rule.EventNotification: one person’s copy of one event, with its in-app, email and SMS delivery state.UserNotificationPreferences: a person’s per-family channels, verified phone and pause.