Host intrusion detection for Linux servers
A host-based intrusion detection system (HIDS) runs on the server itself and looks for signs that someone got in: changed system files, suspicious log lines, unexpected programs and connections. A network IDS watches traffic between machines; a HIDS sees what actually happened on one.
| Tool | What it does | How you run it |
|---|---|---|
| AIDE | File and directory integrity checker: builds a database of file hashes and attributes, then reports what changed | In most distributions; you run checks on a schedule and read the report |
| OSSEC | Log-based intrusion detection, file integrity monitoring, malware detection, active response and compliance auditing | Free and open source, maintained by Atomicorp; agents report to a server you run |
| Wazuh | A broad security platform that grew out of OSSEC: integrity monitoring, vulnerability detection, log analysis, compliance | Free and open source; you run a server, an indexer and a dashboard, or buy Wazuh Cloud |
| Falco | Runtime security that parses Linux system calls and alerts when a rule matches, with container and Kubernetes context | Graduated CNCF project; you write or tune rules and forward its alerts |
| rkhunter | Scans for known rootkits and suspicious file changes | Last release 1.4.6, in February 2018 |
What a HIDS should notice
- Changed files that should not change: system binaries, SSH and sudo configuration, cron and systemd units.
- Logins and failed logins, and who logged in from where.
- Programs and connections nobody expected: a new process listening on a port, an outbound connection from a web server.
- Persistence, the ways an intruder makes sure they come back after a reboot.
What to check before you choose
- Who reads the alerts. A tool that writes a nightly report nobody opens detects nothing in practice.
- What it does on its own. Some only alert; some can block or stop things. Decide what you are comfortable with running unattended, and check that it can be undone.
- What you have to run. A central server and an indexer are a project of their own; a single binary on each host is not.
- Whether it is maintained. Detection is only as good as its last update.
- How noisy it is after a package upgrade or a deploy, the moments when files change for good reasons.
Where Vigilhold fits
Vigilhold is a hosted host security agent for Linux servers and Docker hosts. One command installs the agent; it watches processes, network connections, file changes and logins, and the dashboard is hosted, so there is no server or indexer to run and nothing to tune before it works. It responds only within limits you set, asks first for anything beyond them, and records every action in an audit log, with bans, quarantines and lockdowns you can undo.
Checked in November 2026: AIDE, OSSEC, Wazuh agent docs, Falco docs, Rootkit Hunter.
Vigilhold vs fail2ban · Vigilhold vs CrowdSec · Vigilhold vs Wazuh · Start a 14-day free trial