Notifications
This guide explains how LEA Platform tells you that something happened: the bell, pop-ups, desktop alerts, email, Slack, Teams and webhooks. It covers what each user can tune for themselves and what administrators set up for everyone.
Before you start
- Everybody can use the bell and My account → Notifications.
- You need the Admin role for Settings → Notifications, Settings → Email, Slack & Teams and Settings → Webhooks.
- On a site without internet, email needs a mail server inside your network (a relay). Slack and Microsoft Teams are internet services; see On a site without internet.
How notifications work
A notification is one event: a rule started failing on a host, a host stopped reporting, a backup is overdue, someone changed a setting, and so on. Each notification has:
-
a severity: info, warning or critical;
-
a category:
Category What it covers Compliance Regressions (a rule starts failing), recoveries, bursts and evidence snapshots Fleet Enrollment, hosts that stop reporting, certificates and agent versions Security Blocked commands, identity mismatches, failed sign-ins Platform Cluster health, backups, license, functions and webhooks Changes Who changed rules, groups, users, tokens and settings AI AI assistant approvals and anomalies -
an audience: everyone, administrators only, or only the person it is about (for example "your token expires soon").
If your access is limited to some host groups, you only get notifications about hosts and groups you can see: in the bell, pop-ups, desktop alerts, email, Slack, Teams and your digests.
When many rules flip on one host at once (4 or more), you get one combined notification instead of dozens. When the same change hits many hosts, you get one fleet-wide incident instead of one notification per host; see Fleet-wide incidents. Some notifications are conditions (for example "backup overdue"): they stay open until the problem is gone, then show RESOLVED.
Each person has their own read state, preferences and mutes. They are stored on the server, so the unread count is the same in every browser and device.
Fleet-wide incidents
A fleet-wide incident is one notification that stands for the same change on many hosts, for example "CIS 1.1.2 fails on 1,987 of 2,010 hosts".
It covers a rule that fails or passes again on many hosts, many hosts becoming non-compliant or compliant again, remediation of one rule failing on many hosts, and many hosts that stop reporting, have expiring or expired agent certificates, wait to be activated or enroll at once.
- It starts once enough hosts are affected (10 by default, within 30 minutes). Notifications that arrived for those hosts before that are folded into it and leave the panel and the unread count.
- While it is open it keeps growing. When more hosts join, it moves back to the top and becomes unread again, at most every 10 minutes by default.
- It shows RESOLVED once fewer hosts than the minimum are still affected, or, for one-off events such as enrollments, after 30 quiet minutes.
- Click Show … hosts on it to see up to 100 of the affected hosts.
| Label | What it means |
|---|---|
| FLEET-WIDE | Most of the fleet changed at once, which usually means a rule or fleet-wide change, not separate problems on each host. |
| … hosts · …% of fleet | How many hosts the incident covers, and their share of all hosts. |
Administrators set the numbers in Fleet-wide grouping.
Use the bell and the panel
The bell is at the top right of every page, next to your name. A number on it is your unread count.
-
Click the bell, or press
Shift+N.You should see the Notifications panel. At the top, a small label shows the connection: Live (new notifications arrive instantly), Polling (checking every 15 seconds) or Connecting.
-
Use the tabs to filter: All, Unread, or one category (Compliance, Fleet, …). A number on a tab is its unread count.
-
Click a notification to go to its page: the host, rule, result, group or settings page it is about. Your command finished opens the host's Terminal tab; a started debug capture opens its Logs tab.
-
Use the buttons on each notification:
Button What it does Mark as read / Mark as unread Changes the read state of this one notification Mute… Silences this kind of notification, a host, a rule, a group or a whole category. See Mute a notification Hand icon ("Take ownership so others know it is handled") Administrators, warning and critical only: marks it OWNED BY <you> for everybody. Click again to release More Open, Open rule versions and Mute this rule (fleet-wide incidents about a rule), View in audit log (administrators, for changes) and Copy text -
Click the double tick ("Mark all as read",
Shift+A) to mark everything in the current tab as read.
At the bottom of the panel:
- Show muted shows muted notifications again (marked MUTED).
- Export downloads the last 30 days as a CSV file.
The panel loads more notifications as you scroll, up to the latest 300. For older ones, use Export.
Labels you may see on a notification: RESOLVED (the problem is gone), FLAPPED 2× (it came back after it was resolved), FLEET-WIDE (most of the fleet changed at once), OWNED BY …, FOR YOU (only you received it) and MUTED.
Keyboard keys inside the panel: j/k move, Enter opens, o opens the page, x toggles read, m mutes.
Note: If the panel shows "Notifications may be delayed: no node is running the notification producer.", the server service that creates notifications is not running. See If something goes wrong.
Pop-ups and desktop alerts
- A pop-up appears at the top right of the console when a notification arrives. It has View, Mark read and Mute 8h, and an × to dismiss it.
- A desktop alert is a message from your operating system. It appears when the LEA Platform tab is in the background or minimised.
To turn on desktop alerts:
-
Open My account → Notifications.
-
Under Desktop alerts and testing, click Allow desktop alerts, then allow it in the browser's question.
You should see DESKTOP ALERTS ALLOWED. If you see BLOCKED IN BROWSER SETTINGS, allow notifications for this site in your browser settings.
-
Click Test info, Test warning or Test critical to see how each severity looks.
Pause pop-ups
Pausing stops pop-ups and desktop alerts for a while; the panel still collects every notification. Open the panel, click the pause icon ("Pause pop-ups") and pick 1 hour, 8 hours, 24 hours or Until tomorrow 8:00. The panel then shows "Pop-ups paused until …. The inbox still collects everything."
| Button | What it does |
|---|---|
| Resume | Ends the pause now, so new notifications pop up and show desktop alerts again. Resume now in the pause menu does the same. |
Choose what reaches you
All personal settings are in My account → Notifications. Changes take effect when you click Save at the top right; Discard undoes them.
Pick a profile
Under Start from a profile, click the profile that fits your job. A profile is a ready-made set of choices; you can fine-tune it afterwards.
| Profile | Meant for |
|---|---|
| Operator | Everything that needs hands on the platform: regressions, fleet, security and platform health. Default for administrators |
| Security / CISO | Security events, critical regressions, posture drops and changes to users, roles and keys |
| Auditor | Configuration changes, evidence snapshots and security events, without pop-ups |
| Executive | Posture trends, license and critical security only |
| Observer | Compliance regressions, recoveries and posture. Default for read-only users |
When you change anything below, the profile shows Customized. Click the × on Customized to go back to the plain profile.
Set delivery per channel
Under Delivery, the table has one row per category and one column per channel: In-app (the bell), Pop-up, Desktop, Email, Slack and Teams. In each cell pick the lowest severity that reaches you there:
| Choice | Meaning |
|---|---|
| Everything | Info, warning and critical |
| Warning and up | Warning and critical |
| Critical only | Critical only |
| Off | Nothing on this channel |
A column marked NOT CONNECTED needs your address first. Click Connect email, Slack or Teams. See Connect your own email, Slack or Teams.
Quiet hours
- Under Quiet hours, turn on Enable quiet hours.
- Set From and To (default 22:00 to 07:00, in your browser's time zone).
- Click Save.
During quiet hours there are no pop-ups or desktop alerts. Email, Slack and Teams messages wait and arrive as one summary when the quiet hours end. Critical alerts still get through unless you turn that off under My account → Email, Slack & Teams.
Group notifications by rule
Grouping by rule shows notifications about the same rule that follow each other in the panel as one row, for example "CIS 1.1.2 failed on 4 hosts".
- Under Panel, turn on Group notifications for the same rule (on by default).
- Set Group from to how many notifications in a row make a group (2 to 50, default 3).
- Click Save.
Click a group row to show or hide its notifications. The tick on the row marks all of them as read. This applies to rule failures, recoveries and failed remediations, also older ones; large waves already arrive as one fleet-wide incident.
Single notification types
Under Individual notifications, click Show … notification types to list every type. For any type and channel, pick On or Off to override your delivery settings, or Default to follow them.
Mute a notification
Muting stops something from raising the badge, popping up or reaching your desktop, email, Slack or Teams. Muted notifications stay in history.
- In the panel, click Mute… on a notification.
- Pick what to mute: This kind of notification, Everything about <host>, This kind on <host>, Everything about <rule>, Everything about <group>, or All <category> notifications.
- Pick For how long?: 1 hour, 8 hours, 24 hours, 7 days or For good.
To see or end mutes, open My account → Muted. Click Change to set a new duration or Unmute to end it.
| Column or button | What it means |
|---|---|
| Muted | What you silenced: a notification type, a category, or everything about one host, rule or group, alone or combined. |
| Expires | How long the mute still lasts, or FOR GOOD for a mute that stays until you click Unmute. |
| Unmute | Ends this mute now, so these notifications raise the badge, pop up and reach your channels again. |
Connect your own email, Slack or Teams
Open My account → Email, Slack & Teams. Each card shows CONNECTED, NOT CONNECTED or NOT AVAILABLE (your administrator has not set up that service yet).
- SSO users get their directory email automatically ("from your single sign-on directory").
- Otherwise type your address in Email address, click Verify, type the code from the email into 6-digit code from the email and click Confirm. The code is valid for 15 minutes.
- Click Send test to check it.
Slack: click Connect Slack. LEA Platform finds your Slack account by your verified email. If your Slack email is different, type /lea link in Slack and follow the link.
Teams: in Teams, add the LEA Platform app (your administrator published it), say hi, then come back and refresh.
Further down:
- Digest: pick a Frequency (Off, Every day, Every week), the day and hour, the channels, and LEA AI summary.
- Behavior: hold messages during quiet hours, let critical alerts break through, and get told when a problem is resolved.
Set notification defaults for everyone
Administrators set these in Settings → Notifications. Click Save at the top right of the section when done.
- Under Default profile per role, pick the profile new Administrators and Read-only users start with. People can still pick their own under My account.
- Under Notification types, turn the switch off for a type to stop it for everyone. The table shows each type's Category, Severity and Who sees it.
| Column | What it shows |
|---|---|
| On | Whether LEA Platform creates this type of notification at all. Off stops it for everyone, in the console and on every channel. |
| Who sees it | Which people receive this type: everyone, administrators only, or only the person it is about (for example the owner of an expiring token). |
How long notifications are kept is set in Settings → Data retention (Notification Retention (Days)). Open alerts are kept until they resolve.
Fleet-wide grouping
Fleet-wide grouping turns many notifications about the same change into one fleet-wide incident for everyone.
-
Open Settings → Notifications.
-
Under Fleet-wide grouping, leave Group fleet-wide changes into incidents on. Turned off, every host gets its own notification again.
-
Adjust the numbers if needed:
Field What it sets Minimum hosts How many hosts must change the same way before their notifications become one incident. Default 10. Window (minutes) How far back earlier notifications count towards the minimum, and how long a quiet one-off incident stays open. Default 30. Re-alert interval (minutes) How often a growing incident may move back to the top as unread. Default 10; 0 means every time it grows. Fleet-wide at (%) Share of hosts at which an incident is marked FLEET-WIDE. Default 50. -
Below the numbers, turn off any kind of incident you do not want; those changes then arrive one host at a time.
-
Click Save.
Changes apply within a minute. The incident types also appear under Notification types; turning one off there also brings back one notification per host.
Send notifications by email, Slack and Teams
Administrators set this up in Settings → Email, Slack & Teams. The section has five tabs: Connections, Routes, Delivery log, Recipients and Options. The strip at the top shows connections, failing connections, routes, and messages sent and failed in the last 24 hours.
Step 1: set the public address
- Click the Options tab.
- Under Public address, type the address people use to open LEA Platform in Public URL of LEA Platform, for example
https://lea.example.com. If a Use https://… button appears with the right address, click it. - Click Save.
Without it, messages have no links, and Slack and Teams buttons cannot reach LEA Platform.
Step 2: connect email
Connect email first: it also delivers verification codes and digests.
-
Click the Connections tab, then Connect Email.
-
In Provider, pick how LEA Platform sends mail: SMTP server, Microsoft 365 (Graph API), Google Workspace (Gmail API), Amazon SES, SendGrid, Mailgun or Postmark.
-
Give it a Name, for example
Corporate mail. -
For SMTP server, pick a Preset to fill in the server settings. For an internal mail server, pick On-premises Exchange / internal relay and type its address in SMTP server.
-
Fill in From address (required), From name and optionally Reply-To (optional).
-
Fill in the provider fields. For SMTP:
Field What to enter SMTP server Host name of the mail server, for example smtp.corp.localPort 587for STARTTLS,465for SSL/TLS,25for many internal relaysEncryption STARTTLS (587), SSL/TLS (465) or None (trusted internal relay only) Authentication User name and password, OAuth 2.0 for Microsoft 365 or Google Workspace, or None (IP-allowed relay) User name / mailbox, Password or app password Only with password authentication -
If your mail server uses a certificate from your own certificate authority, open Advanced: proxy, certificates, endpoints and paste it into Trusted CA certificate (PEM, optional).
-
Leave Enabled on. Turn on Default for personal emails and verification codes.
-
Click Connect.
-
On the new connection card, click Check (signs in without sending), then Send test and type your address in Send to.
You should see CONNECTED on the card.
Warning: Do not turn off Verify TLS certificates to make an error go away. Anyone on the network path could then read or change messages. Add your CA certificate instead.
Step 3: connect Slack or Teams (optional)
Click Connect Slack or Connect Microsoft Teams on the Connections tab and pick a provider:
| Provider | What you get |
|---|---|
| Slack app (bot token) | Buttons, direct messages to people, /lea commands, messages updated in place |
| Slack incoming webhook | Posts to one channel, links only |
| Teams bot (Azure Bot) | Adaptive Cards with buttons, personal chats, channel messages, messages updated in place |
| Teams workflow webhook | Posts to one channel, links only |
The dialog lists the setup steps under How to set it up. For a Slack app, save it first as a draft (Save draft), then click Create Slack app on the card: it opens Slack with every URL and permission prefilled. For a Teams bot, download the Teams app package from the card and upload it in the Teams admin center.
Step 4: create a route for a team
A route sends shared notifications to a team: a Slack channel, a Teams channel or group chat, or a list of email addresses.
- Click the Routes tab, then New route.
- Type a Name, for example
SOC critical alerts. - In Send with, pick the connection.
- Under Destination, pick the Channel (Slack), the Team channel or group chat (Teams), or type Recipients separated by commas (email). Each email recipient gets their own copy with an unsubscribe link.
- Under What to send, set Minimum severity and, if you want, limit it to some groups, categories (click the chips), notification types or frameworks.
- Under When, pick As it happens, Scheduled digest or Both. For a digest, set Every (Day or Week), On (weekly), At and Time zone.
- Under Options, choose follow-ups when a problem is resolved, an executive summary by LEA AI at the top of the digest, and for Slack Mention on critical alerts.
- Leave Route enabled on and click Create route.
- In the route's row, click the send icon ("Send a test message"). For a digest, the clock icon sends one now ("Send a digest now").
Personal notifications (for example "your token expires") never go to routes.
| Column | What it shows |
|---|---|
| On | Whether the route is active. A route that is off sends nothing. |
| Route | The route's name, with what it sends below it: the minimum severity, the categories and any type or group filters. |
| When | When the route sends: each notification as it happens, a scheduled digest at the time shown, or both. |
Options worth knowing
On the Options tab:
| Option | What it does |
|---|---|
| Detail in messages | Full (details, facts, AI analysis) or Minimal (title and link only). Minimal keeps host names, output and AI text inside LEA Platform |
| Product name in messages | The name shown as sender in messages |
| Send a follow-up when a problem is resolved | Sends a short "resolved" message |
| Group into one message from | When this many messages wait for one destination, they become one summary |
| Skip events older than | Events older than this (in minutes) are not sent after an outage |
| Stop for users inactive for | People who have not signed in for this many days stop getting messages (0 = never) |
| Add AI analysis to alerts | Adds the cause and fix from the AI assistant to serious alerts. For sets the severity, At most per hour limits the number. Needs the AI assistant |
| HTTP proxy | Outbound proxy for Slack, Teams and email APIs. Each connection can override it |
| Keep the delivery log for | Days of delivery history |
Check what was sent
- The Delivery log tab lists every email, Slack and Teams message with its status: SENT, QUEUED, SENDING, RETRYING, FAILED or SKIPPED. Filter by Search, Status and Connection. Click the retry icon on one message, or Retry failed for all.
- The Recipients tab lists everyone who connected an address. Turn off the switch in the Delivery column to stop all email, Slack and Teams messages to that person.
| Column or button | What it means |
|---|---|
| Retry failed | Queues every RETRYING or FAILED message of the last 3 days again, for the connection picked in Connection or for all. |
| Digest | How often this person gets a summary of their notifications by email, Slack or Teams. A dash means no digest. |
| Delivery | Whether email, Slack and Teams messages may reach this person. Off stops all of them; the bell in the console still works. |
Failed messages are retried with growing delays for up to 8 attempts. After three failures in a row (or one sign-in failure) the connection is marked FAILING, and a "delivery failed" notification goes out through the other channels.
Connection status
Each connection card on the Connections tab shows one status:
| Status | What it means |
|---|---|
| CONNECTED | The last check or message through this connection worked. |
| NOT TESTED | No check or message has gone through this connection yet, so it is not known whether it works. Click Check or Send test. |
| FAILING | Messages through this connection keep failing. The card shows the last error. |
| RE-ENTER CREDENTIALS | LEA Platform can no longer decrypt this connection's stored password or token, usually because the server secret changed. Edit the connection and enter it again. |
| DISABLED | The connection is switched off and sends nothing. |
Use chat commands in Slack and Teams
With a Slack app or Teams bot connected, people can ask LEA Platform from chat. In Slack type /lea <command>; in Teams send the command to the LEA Platform app or @mention it in a channel.
| Command | What it does |
|---|---|
status | Compliance posture right now |
host <name> | One host: score and failing rules |
rule <name> | One rule: where it fails |
ask <question> | Ask LEA AI about your environment (you can also just write the question) |
subscribe [critical|warning|all] [categories] | Administrators: post alerts in this channel (creates a route) |
unsubscribe | Administrators: stop alerts in this channel |
link / unlink | Connect or disconnect your chat account and your LEA account |
whoami | Shows which LEA account and role you are connected as |
Example, posting critical compliance and security alerts to the current Slack channel:
/lea subscribe critical compliance security
Buttons and commands act as the linked LEA user, with that user's current role. People whose access is limited to some host groups get answers about those hosts only. Acknowledge works for operators and administrators, on notifications they are allowed to handle. The AI in chat only reads; it never makes changes. When someone's LEA account is disabled or removed, chat commands stop working for them and they no longer get messages or digests.
Answers with LEA data stay private to the person who asked:
- Slack: in a channel, answers to
/lea, @mentions and Explain with AI are shown only to you ("Only visible to you"). Direct messages with the app are unchanged. - Teams: in a channel or group chat,
status,host,rule, questions and Explain with AI get a short reply asking you to message LEA Platform directly. Ask again in your personal chat with the app.
Administrators who want answers posted for the whole channel, as before, can turn on Post answers with LEA data in channels in the Slack or Teams connection's advanced options (API: PATCH /lea/integrations/<id> with {"config": {"channel_answers": true}}).
Limit what a channel can see
When answers are posted in channels, everyone in the channel reads them. An administrator can give a channel its own role and host groups, for example Viewer on Linux-Prod for #linux-team. Answers posted in that channel then show only what both the person who asked and the channel may see: a channel never shows more than the person asking, and never more than its own access. If the two have nothing in common, LEA Platform answers that the channel's access does not include the data; ask in a direct message instead.
Channels without their own access behave as before. Private answers ("Only visible to you", personal chats) always use your own access.
To give a channel its own access (needs the Integrations permission):
- Open Settings → Email, Slack & Teams, tab Connections.
- On the Slack app or Teams bot connection, click Channel access.
- Find the channel in the list (use Search channels) and click it, or its pencil icon.
- Pick a Role and, to limit it to some host groups, a Scope. Click Save. If your organization asks for it, confirm it is you.
You should see the role and groups next to the channel. Channels with their own access are listed first.
| Column | What it shows |
|---|---|
| Role | The most a channel's posted answers may show; Not limited when the channel uses the access of the person asking. |
| Scope | The host groups the channel's answers are limited to, or Everything. |
To go back to the person's own access, click the bin icon next to the channel and Remove. A channel that LEA Platform can no longer list (for example, the app was removed from it) stays in the list while it has its own access, so you can remove it.
Channel access only matters while Post answers with LEA data in channels is on for the connection; the dialog says so when it is off. Slack private channels appear after you invite the app; Teams channels appear after the LEA app is added to the team.
Approve requests from Slack and Teams
When someone needs your approval, the Approval requested message in your Slack direct messages or your personal Teams chat has Approve and Reject buttons, if you are allowed to decide the request. Clicking one decides it as your LEA account, exactly as in the console, and the message changes to show who decided and when, for example "Approved by Dana and done".
Some requests can only be decided in the console. Their message has Open in LEA instead and says why:
- the action needs a sensitive (privileged) permission, for example running commands on hosts;
- it is a request for temporary access;
- your organization asks approvers to confirm their identity before approving (the Approvals re-authentication setting in Settings → Security); chat cannot ask you to sign in again.
To get these messages in chat, turn on Security notifications for Slack or Teams in My account → Notifications, and connect your chat account. A click from a chat account that is not connected answers with the link to connect it.
Send compliance state to another system with webhooks
A webhook tells another system — NAC, firewall, ticketing, SIEM or a CMDB — when a host, a rule or a result changes. Set it up under Settings → Webhooks; every step, from presets to dry runs and the log, is in Webhooks.
Note: Webhooks are not a general alert channel. For alerts to people, use email, Slack or Teams routes.
On a site without internet
| Channel | Works offline? | What you need |
|---|---|---|
| Bell, pop-ups, desktop alerts | Yes | Nothing |
| Yes, with an internal mail server | An SMTP relay inside your network. Use the preset On-premises Exchange / internal relay, and paste your CA under Advanced if it uses your own certificates | |
| Webhooks | Yes, to internal systems | The receiving system must be reachable from the LEA server |
| Slack, Microsoft Teams | Only with a controlled path to the internet | Slack and Teams are internet services. LEA Platform must reach them through HTTP proxy, and for buttons and commands Slack or Teams must reach your Public URL of LEA Platform. Without that, do not connect them |
| AI analysis in messages | Only with a local AI model | See AI assistant |
More in Running without internet.
If something goes wrong
| Symptom | Fix |
|---|---|
| The bell shows Polling instead of Live | A proxy or firewall between your browser and LEA Platform blocks long-lived connections. Notifications still arrive every 15 seconds |
| "Notifications may be delayed: no node is running the notification producer." | The background services are down. On the server run docker compose ps in the release directory and start the stack with docker compose up -d |
| No pop-ups at all | Check the pause icon in the panel, your quiet hours, and the Pop-up column under Delivery |
| Desktop alerts never appear | Your browser blocks them (BLOCKED IN BROWSER SETTINGS). Allow notifications for this site |
| An email card says NOT AVAILABLE | No email connection is enabled. An administrator must connect one |
| A connection shows FAILING | Read the red message on the card, then click Check. Look at the Delivery log for the exact error |
| A connection shows RE-ENTER CREDENTIALS | The server secret changed. Edit the connection and type the password or key again |
| Messages have no links | Set Public URL of LEA Platform on the Options tab |
| A webhook does nothing | See Webhooks → If something goes wrong |
More in Troubleshooting and the notifications reference.