Every time you open it, Security Insights reads your authentication, privilege and system logs and runs fifteen checks over them. Did anyone get in who shouldn't have? Did an attack work? Did someone add an account, install software or wipe a log? You get an answer for each check, including the ones that came back clean.
Most security dashboards show you a count of events and leave the interpretation to you. This page starts from the questions a sysadmin actually asks after a weekend away:
Each check is either flagged or reported clean. The right-hand column is the exact sentence you see when nothing was found.
| Check | What it looks for | When it's clean |
|---|---|---|
| Successful login from Tor | Someone signed in through a Tor exit node | No successful logins came through Tor. |
| Successful login from a blocklisted address | Someone signed in from an address on a known-bad list | No successful logins came from blocklisted addresses. |
| SSH login as root | Direct root logins over SSH, rather than sudo from a named account | Nobody logged in directly as root over SSH. |
| Brute force succeeded | Repeated failures from an address, then a login that worked from that same address | No authentication attack was followed by a successful login. |
| Brute force attempts | Repeated failed logins from the same address in a short window | No brute force attempts were detected. |
| SSH login with a password | Password logins over SSH, where keys would be stronger | No SSH logins used a password. |
| Accounts added | New user accounts created on any host | No user accounts were added. |
| Accounts deleted | User accounts removed from any host | No user accounts were deleted. |
| Failed sudo | Someone tried to elevate with sudo and got the password wrong | No sudo authentication failed. |
| Failed su | Someone tried to switch user and got the password wrong | No su authentication failed. |
| Applications installed | New software appearing on a host | No new applications were installed. |
| Event log cleared | Audit logs wiped, often the first thing an intruder does | No event logs were cleared. |
| Disk filling up | Warnings that a disk is full or nearly full | No warnings about disk space. |
| Low memory | Out-of-memory warnings and kills | No warnings about low memory. |
| System crash | Logs that point to a crash | No crashes were reported. |
There is also a review list of every SSH login, by user and host. Nothing in it is flagged, but it's the quickest way to see who has been on which machine.
A flagged check comes with the evidence already pulled together. You get the top five of each:
Under each finding, See these logs runs the search behind it. You go straight from "three password logins over SSH" to the three log lines, then from any address in them to everything else that address has done across your servers.
An empty dashboard and a clean one look the same, and only one of them tells you anything. Security Insights lists every check that came back clean, in a plain sentence, under Checked, nothing found.
That matters for two reasons. First, you can see the checks actually ran, so a quiet page doesn't mean a broken one. Second, it's the evidence an assessor asks for. "We monitor for failed privilege escalation and recorded none this quarter" is a complete answer under PCI DSS, provided you can show you were watching. Our PCI DSS Requirement 10 guide covers this in more detail.
The checks are only as good as what feeds them. Every log is parsed and categorized as it arrives, using detection logic from the people who wrote OSSEC.
Detection is useful. Not having to deal with the same address again is better.
Turn on Active Response and every address that triggers a brute force, a web scan or a burst of server errors goes onto a plain-text IP blocklist unique to your account. New addresses appear within seconds. Each one stays on the list for as long as you choose, from ten minutes to two days (the default is thirty minutes), and then drops off. Private, reserved and carrier-grade NAT ranges are never added, so a misbehaving internal host can't lock out your own network.
Trunc publishes the list; your firewall enforces it. You pull the list on a schedule and load it into whatever does the blocking. For Linux, our ipset script downloads it and swaps it into an iptables set atomically. Anything that can fetch a URL and read one IPv4 address per line can use it. How Active Response works.
Treat the list URL like a credential. It's long and unguessable, but anyone who has it can read it.
Reading the page is for when you sit down to look. Alerts are for when you don't. The events behind these checks have default alerts you can turn on per destination:
Alerts go to email addresses and Slack webhooks, as many of each as you like, and you choose which alert goes where. You can also write your own rules, matched against every incoming log and scoped by category. To keep an attack from becoming an inbox flood, alerts are capped at a per-hour limit you set. Alerts and notifications help.
Security Insights works from:
The supported logs page has the full list.
Some things it doesn't do, so you can plan around them:
Do I need to configure anything?
No. Once logs from a Linux or Windows host are arriving, the checks run on them. The only settings are optional: brute force sensitivity, and whether to turn on Active Response.
Is Security Insights included in my plan?
Yes. Security, Web and Operational Insights are part of every plan, including Basic. See pricing.
What period does it cover?
You choose the period at the top of the page, from today back across the last several weeks.
Does a flagged check mean I've been breached?
Not necessarily. "SSH login with a password" is a hardening recommendation. "Brute force attempts" means someone tried. "Brute force succeeded" and "Successful login from Tor" deserve your attention today. The page groups findings into Needs attention and For review so the urgent ones come first.
Can it block attacks automatically?
It can publish the attacking addresses for your firewall to block. See Active Response above. Trunc doesn't sit in front of your servers, so blocking depends on your firewall pulling the list.
14 days free trial. No credit card required.