CIS-aligned security automation for Debian, RHEL, Arch, SUSE, Alpine, Void, Gentoo and NixOS, plus macOS and the BSDs. 23 hardening modules, a preview mode that shows every change before it happens, and a full revert path for when something goes wrong.
See the real screenshots or what Pro adds (coming soon).
Every module is individually selectable, risk-labelled, and shows a live
[enable ] / [disable] status
derived from the compliance checks, so you can see what is already in place
before you touch anything.
Root login, password auth, empty passwords, X11 and TCP forwarding, MaxAuthTries, login grace and keepalives — validated with sshd -t before restart. Plus GRUB password protection.
nftables, UFW or firewalld firewall, Fail2Ban, sysctl network hardening, unused protocol blacklisting, remote syslog.
auditd rule sets, AIDE file integrity, rkhunter rootkit scanning, etckeeper change tracking.
ASLR, hardlink/symlink protection, SUID/SGID reduction with per-binary backup, UMASK 027 defaults.
AppArmor profile enforcement on Debian family, SELinux enforcing mode on RHEL family.
Docker daemon hardening, ModSecurity WAF, USB storage blocking, compiler access restriction.
The module states and check results are demo values β the layout, ordering, risk levels, remediation hints and score arithmetic are the real thing. To click through every menu choice with output recorded from a real run, try the interactive replay.
Eight Linux families and five Unix/BSD systems, or auto-detect.
Read-only. Every failure names the module that fixes it.
This script rewrites SSH config, firewall rules and file permissions on your machine. That is a lot of trust to ask for, so every destructive action has a way back.
--auto-mode --dry-run prints every change the script would make and modifies nothing; --dry-run alone previews from the menu. Run it first, every time.
Modified files are copied to a timestamped /root/hardening_backup_* directory before any edit, including original SUID/SGID permission modes.
--revert restores from backup; --revert-suid rolls back only permission changes. Revert auto-finds the most recent backup.
Password policy is set in /etc/login.defs only. Nothing under /etc/pam.d is ever modified.
Nothing is ever sent anywhere about your hosts. In the free tier the only outbound traffic is the optional, explicitly-invoked update check against the GitHub API, which asks for the latest release and this project's published security advisories. The paid tiers can additionally download public vulnerability data β your distro's security feed, and the CISA KEV and EPSS lists β but only when you pass --cve-feeds or --cve-enrich, which are off by default. Those are inbound fetches of public files; no query carries anything identifying. Enterprise GitOps fetches policy only from the git repository you name.
Plain Bash, MIT licensed, ShellCheck-clean in CI on every push, tested in fresh containers on 10 distros across the Debian, RHEL, Fedora, Arch and Alpine families.
/etc/pam.d/common-password directly. On a Mint box
running KDE Plasma, a later pam-auth-update run reconciled around that
untracked edit and broke the SDDM login. The fix was to stop modifying PAM altogether, which is
why the current release sets password policy in /etc/login.defs instead.
Being precise here matters more than sounding impressive, so here is exactly what each claim means.
Built-in checks follow CIS Benchmark guidance for Linux. This is alignment, not a CIS certification, and the built-in scorer is not an accredited assessment.
Runs oscap against the official SCAP Security Guide, so you get a real, assessable result under a CIS or DISA STIG profile β not the tool's own interpretation of one.
Maps each hardening module to the NIST SP 800-53 Rev. 5 controls it contributes to, and reports how many are in place on this host right now. Exports to JSON and CSV, and appears in the HTML compliance report. A mapping and a configuration record β not an assessment, and not a claim of compliance.
What costs money is everything around the hardening: proving the work happened, repeating it on a schedule, and doing it across more machines than you can log into by hand. One server and an evening? The free tier's 23 modules are the same engine the paid tiers run on.
| Capability | Free | Pro (coming soon) | Enterprise (coming soon) |
|---|---|---|---|
| Hardening modules | 23 | 23 | 23 |
| Dry-run & full revert | β | β | β |
| Built-in CIS checks | β | β | β |
CIS results as JSON / SARIF (--json, --sarif) | β | β | β |
Skip single settings (UH_SKIP_SETTINGS) | β | β | β |
| One-step fix for each failed CIS check (Free applies all 23 modules) | β | β | β |
| HTML compliance report | β | β | β |
| Posture: ports, logins, services | β | β | β |
| Extra hardening modules (SSH rate limiting, sshd resource caps, USB device control) | β | 3 | 3 |
| Exception register (waivers) | β | β | β |
| Intrusion activity & new-ban alerts | β | β | β |
| Compromise indicator scan, baseline & scheduled email | β | β | β |
| Asset inventory & SBOM export | β | β | β |
| CVE exposure report | β | β | β |
| AI directory lock-down | β | β | β |
| Scheduled runs & email alerts | β | β | β |
| OpenSCAP CIS / STIG scan | β | β | β |
| NIST 800-53 crosswalk | β | β | β |
| Policy-as-code & drift detection | β | β | β |
| GitOps delivery, signed approvals, host tags | β | β | β |
| Multi-host dashboard | β | β | β |
| Remote SSH deployment | β | β | β |
| Fleet compromise-scan view & remote scan | β | β | β |
| Run history export (JSON / CSV) | β | β | β |
| Remote syslog forwarding | β | β | β |
| Fleet-wide syslog (one collector, pushed on deploy) | β | β | β |
| Support | Community | Priority |
The 23 modules in Pro and Enterprise are the same code the free tier runs. Nothing is removed or crippled to make the paid tiers look better β open the Pro menu above and all 23 free modules are sitting there above the paid panel. What you buy is the operational layer: evidence, repetition and scale, plus a few extra hardening modules. It also means the free tier cannot be quietly degraded to drive upgrades, because the paid tiers are built on it.
The Pro/Enterprise licence ends at the end of the paid period and you stop
running that software. The free tier keeps working under MIT, unaffected.
Hardening already applied stays applied, and --revert
still works β both live in the free tier.
--dry-run first, and
--safe-only to apply everything except those two.
It can β any real hardening tool can, which is why the safety design matters.
Run --dry-run first to see every change, keep the automatic backups,
and use --revert if something misbehaves. The two High-risk modules
(SUID stripping, GRUB password) are labelled: the menu asks before applying them, and
--safe-only leaves them out of an --auto-mode run. Without it,
--auto-mode includes SUID stripping; the GRUB password always needs you at the keyboard.
SSH hardening disables root login and password authentication. If you connect as root with a
password and no key, you will lock yourself out. Set up key-based access for a non-root sudo user
first, and preview with --dry-run before applying.
No, and anything claiming otherwise deserves scrutiny. The built-in checks are CIS-aligned. For an assessable, auditor-recognised CIS or DISA STIG result, the Enterprise tier runs OpenSCAP against the official SCAP Security Guide content.
Eight Linux families: Debian/Ubuntu/Mint/Kali/Raspbian, RHEL/CentOS/Fedora/Rocky/AlmaLinux/Amazon,
Arch/Manjaro/EndeavourOS/Artix (Artix untested), openSUSE/SLES, Alpine, Void, Gentoo and NixOS.
Artix is treated as Arch but runs without systemd, so it has not been tested. FreeBSD,
OpenBSD, NetBSD and Solaris are detected and adapted, with thinner coverage. On macOS it runs the
read-only checks and the dry run only; applying hardening there is not supported yet (it needs bash 4
or newer: brew install bash). The script detects your distro or lets you select it, and adapts package manager and service
names accordingly.
No. There is no telemetry, analytics or licence phone-home, in any tier β nothing about your hosts leaves them. In the free tier the only outbound request is the optional update check you invoke yourself from the menu, which asks the public GitHub API for the latest release and for this project's published security advisories. Everything else is local.
The Pro and Enterprise CVE features default to your own package manager's repository
metadata, so they need no new network access at all. Two optional flags will download public
files if you ask for them β --cve-feeds for your distro's security
feed, --cve-enrich for the CISA KEV catalogue and EPSS scores. Both
are off unless passed, and both are plain downloads: nothing is uploaded, and the request
contains no host identifier, package list or licence key.
Enterprise GitOps (--gitops) clones and fetches policy from the git
repository you point it at, and nowhere else.
The free tier is MIT-licensed β source is on GitHub.
Enterprise is coming soon. A host is any machine you scan, harden or track in the dashboard. The base tier covers 10 hosts, additional hosts are $3/month each, and the flat $149/month plan covers up to 50. Above that, get in touch and I will size it.
Pro is coming soon; once it launches, get in touch and I will set up a trial. The free tier is also a genuine product rather than a crippled demo, so you can validate the hardening itself before paying for the automation layer.
Clone it, preview every change it would make, and decide from there. It costs you one command.