Wazuh
Wazuh watches every endpoint: file changes, logins, processes, registry keys, vulnerabilities. It is an open-source HIDS/XDR that turns "something changed on that server" into an alert with a rule ID.
What it is
Wazuh is a host-based intrusion detection system that grew into an extended detection and response platform. An agent runs on each endpoint, collecting log data, file hashes, running processes and configuration state. The manager receives that data, decodes it, and applies decoders and rules. The indexer (OpenSearch) stores it and the dashboard visualises it.
Three capabilities matter most in practice: FIM (file integrity monitoring) alerts on changes to files you care about, log analysis applies rules to syslog, Windows events and application logs, and SCA (security configuration assessment) scores the host against CIS benchmarks.
How to use Wazuh
Follow it exactly as you would in a real session — every command, every line of output, and what each result actually tells you. This is a interface walkthrough.
Confirm the agent is actually reporting
Configure file integrity monitoring on what matters
Read a real FIM alert
Search across agents for the same file
Flags & options
| FLAG | WHAT IT DOES |
|---|---|
| rule.level >= 10 | Start with high-severity alerts only. |
| rule.id:550 | File integrity: checksum changed. |
| rule.id:553 | File integrity: file deleted. |
| rule.id:5712 | SSH brute force detected. |
| syscheck.path | Filter by watched path. |
| data.srcip | Filter by source IP. |
| agent.name | Filter by host. |
| Active response | Automatic block or isolate — test in a lab first. |
| SCA | CIS benchmark scoring per host. |
Why it matters
Network sensors miss what happens inside an encrypted connection. Wazuh sees the host: a new file in /var/www/html, a user added to sudoers, a service restarted at 3 AM. For a small team without a commercial EDR, it is the highest-value defensive deployment available — and it is free.
How it works
Each agent collects three kinds of data: log lines, file hashes for watched paths, and system state. It ships them to the manager, which runs decoders to pull fields out and rules to decide whether the event matters. Each alert carries a rule ID and a severity level.
File integrity monitoring works by hashing every watched file on a schedule (and in real time where supported). When a hash changes, the manager compares it against the baseline and raises rule 550 with the old and new checksums.
Commands & syntax
# /var/ossec/etc/ossec.conf on the agent — watch what matters
<syscheck>
<directories check_all="yes" report_changes="yes">/var/www/html</directories>
<directories check_all="yes">/etc</directories>
<directories check_all="yes">/root/.ssh</directories>
<frequency>300</frequency>
</syscheck># in the dashboard, or directly in the indexer rule.level >= 10 # authentication failures rule.id: (550 OR 553) AND data.srcip: * # file added in the web root rule.id: 550 AND syscheck.path: /var/www/html/*
# on the manager /var/ossec/bin/agent_control -l /var/ossec/bin/agent_control -i 001 # on the agent /var/ossec/bin/wazuh-control status
# block an IP for 10 minutes after repeated auth failures
<active-response>
<command>firewall-drop</command>
<location>local</location>
<rules_id>5712</rules_id>
<timeout>600</timeout>
</active-response>root@wazuh-manager:~# grep -i "integrity" /var/ossec/logs/alerts/alerts.log | tail -6 ** Alert 1775000001.88123: - ossec,syscheck, 2026-11-14T09:41:02+05:30 (manager) 10.0.0.5->syscheck Rule: 550 (level 7) - Integrity checksum changed. File: /var/www/html/uploads/shell.php New size: 132 Old md5sum: d41d8cd98f00b204e9800998ecf8427e New md5sum: 9f2c1a7b3e5d4c6a8b0e1f2d3c4b5a69 Mode: realtime # A new PHP file appeared in the uploads directory. # uploads/ + .php + realtime = webshell until proven otherwise. # This is the alert that matters most in a typical web compromise.
Common mistakes
Watching the entire filesystem
You will generate thousands of alerts a day and stop reading them. Watch specific paths.
Alerting at every level
Start at level 10+, tune down once you have a baseline.
Deploying active response without testing
Auto-blocking can lock out administrators or break applications.
Treating Wazuh as a network sensor
It sees the host. Use Suricata or Zeek for traffic.
Ignoring agent health
A disconnected agent is a blind spot. Alert on agents that stop reporting.
No baseline period
Without knowing normal, every alert looks like an incident.
Practice safely
Deploy the Wazuh OVA or the all-in-one installer in a VM, add a second VM as an agent, and trigger your own alerts: change a watched file, fail an SSH login five times, and watch the rule IDs appear. TryHackMe has a Wazuh room.