Notifications and Daily Brief

Notifications and Daily Brief

This configuration screen determines how you want to be notified when Device Sentinel detects a problem. It is divided into three sections and built around a single principle: your phone should only interrupt you when something genuinely needs your attention.

Instant Notifications

These are the live alerts sent the moment a device develops a problem. They follow strict rules to prevent notification fatigue.

The Persistent Card

This is a single card for your Home Assistant dashboard that displays the current trouble state of your entire home. It is organized by kind: a Device line (frozen, unavailable, unknown, or never reported), a Battery line (cells under the threshold and cells running down), and a Signal line (railed links).

A cell under the threshold reads as its level, "Soil Irrigation (Monstera) 0%". A cell running down reads as its forecast, "Door 2nd Bedroom empty in about 2 weeks". A cell that is both reads as one entry with the level first, "Door 2nd Bedroom 16%, empty in about 2 weeks".

It is rewritten every time something changes.

It reads "All devices reporting" when your system is perfectly clean.

Because it is silent by nature, it ignores quiet hours and continuously updates day and night.

You can turn this on or off using the Create a Persistent Notification toggle.

Phone Pushes

Live phone push notifications are sent only to high-priority targets. To prevent spam, pushes are grouped by kind (one for device liveness, one for battery, one for signal); a cell running down belongs to the battery group. When an event occurs, the push tells you exactly what just happened and the current standing state of that kind, giving you the complete picture in one message. For a cell running down the fault line reads "At 5:29 pm, Door 2nd Bedroom battery is running down, empty in about 2 weeks", and drops the forecast where the readings do not support one.

New faults play your phone's normal notification sound.

Recoveries arrive silently. A device coming back online is good news that updates your picture, but it should not interrupt your day.

When a Bridge or Broker Goes Down

Behind a stopped bridge or broker every device goes quiet at once, and a message per casualty buries the one sentence worth reading. So while an upstream is down, its casualties are counted rather than named, on the card, in the pushes, on The Problem List and in the daily brief alike.

An outage is given sixty seconds to settle before anything is sent, because Home Assistant marks devices unavailable over tens of seconds and a count taken at the first one is wrong. The first message about that upstream plays your normal notification sound. Every later one is silent and carries the current tally, so the number on your phone stays true without interrupting you again. The recovery is announced once and silently, carrying how many devices had gone quiet, after the same settle window, so a device that did not come back is named once the others have.

Two devices are never suppressed: one whose problem began before the outage, because it was already broken, and one still down after the upstream returns.

Target Tiers

When selecting your notification services, you assign them to tiers. The difference between these tiers is not loudness, but whether live events reach that person at all.

• High-priority targets: Receive live per-kind phone pushes as they happen, plus a message-sized version of the daily brief.
• Normal-priority targets: Receive only the message-sized daily brief. They are never bothered by live, individual device problems.

If you select a service that no longer exists, the integration logs a service_not_found warning and continues. To fix this, return to this screen and select a valid service.

The Settle Period and Acknowledgment

A device must stay in a fault state for a settle period before any instant notification is triggered, so brief network hiccups do not reach your phone. This period scales to each device's natural rhythm and can be adjusted on the Advanced screen.

If you acknowledge a device in the problem list, it silences all instant notifications about it across all channels: the card, the pushes, and the brief.

If the acknowledged device worsens, nothing is sent, because your acknowledgment was a statement that you are already aware of the issue.

If the device recovers on its own, the recovery is reported, because a recovery is news and the acknowledgment ends when the problem does.

Quiet Hours

During your defined quiet hours, no phone pushes are sent.

If a device drops offline at 3 AM, the event is recorded, the dashboard card updates, and it is added to your daily brief, but your phone will not alert. If a problem opens and heals entirely while you are sleeping, it is noted in the daily brief, but it never triggers a delayed push once quiet hours end.

Daily Brief

This section controls the scheduling and delivery of your daily summary. What the brief actually says and how to read it is covered separately in The Daily Brief.

When to Send It

This single setting does two jobs: it sets the hour the email is delivered, and it defines the 24-hour reporting window.

If you set the brief for 7 AM, the report covers 7 AM to 7 AM, so an overnight network outage is kept together in one logical report rather than split across two days. If you prefer strict calendar days, set this to midnight.

Only the history is bounded by this window. The "what needs attention now" section of the brief is read at the exact moment the file is written, so a 7 AM brief tells you exactly what is broken at 7 AM.

The brief's file is named for the day the window opened. A 7 AM brief sent on the morning of the 24th covers the 23rd through the 24th, so it is named for the 23rd. See The Daily Brief for more details.

How Often

You have three delivery choices:

• Every day: Sends the brief regardless of your system's state. On perfectly clean days, this acts as a reassuring all-clear and proof that the integration is running.
• Only when something needs attention: Stays completely silent on clean days.
• Never: No messages are sent anywhere. The file is still written to your disk and can be viewed manually.

Send the Full Brief To

This is a third target list specifically for services capable of carrying a full, formatted document, most commonly email. High- and normal-priority targets only receive a condensed, message-sized summary; targets in this list receive the complete brief as a styled page, tables included. To use email, configure Home Assistant's SMTP integration once, and its notify service will appear in this dropdown.

Manual Updates

If you press the Regenerate Reports button on the main integration page, the integration immediately rewrites today's working brief. It is marked in progress and covers the last twenty-four hours, replacing itself until the scheduled run closes out the day.