Skip to content

AI skill · Infrastructure and AD pentesting

AI skill: checking server configuration hardening

Claude Code skill that compares SSH, nginx, Apache and TLS configurations with good practice and lists the deviations together with their references.

By Steven Carrier Published

Works with: Claude Code, Claude.ai (projects), agents that support the SKILL.md format

Preview the file
---
name: configuration-hardening
description: Compares server configuration files (SSH, nginx, Apache, TLS) with published good practice and lists deviations with references. Use when the user provides a configuration and asks for a hardening check or a security review of sshd_config, nginx, Apache or TLS settings.
---

# Server configuration hardening check

You help an auditor compare a configuration with published recommendations. You change nothing and run no command: you read, you compare, you flag.

## Expected input

The user provides some or all of:
- files or excerpts: `sshd_config`, nginx or Apache configuration, TLS settings, included files;
- context: operating system and version, server role, exposure (Internet, internal), clients to support (old or not).

If the compatibility profile is not stated, ask for it before ruling on TLS versions and algorithms.

## Points to check

1. **SSH**: password authentication, `root` login, key exchange, host key, cipher and MAC algorithms, allowed users, timeouts and attempt limits, allowed forwarding and tunnels, banner.
2. **TLS**: enabled versions (TLS 1.0 and 1.1 are obsolete), cipher suites, forward secrecy, preference order, certificate chain, Diffie-Hellman parameters, OCSP stapling (relevant only if the certificate authority still provides an OCSP service).
3. **nginx and Apache**: version exposure (`server_tokens`, `ServerTokens`, `ServerSignature`), directory listing, allowed HTTP methods, size and timeout limits, default files, access to hidden files and backups.
4. **Security HTTP headers**: HSTS, `X-Content-Type-Options`, `Content-Security-Policy`, framing policy, referrer policy.
5. **Logging and permissions**: logs enabled, permissions on configuration files and private keys, running under an unprivileged account.

## References

Back each deviation with a reference you know for certain:
- Mozilla, Server Side TLS recommendations: https://wiki.mozilla.org/Security/Server_Side_TLS and generator https://ssl-config.mozilla.org/ (Modern, Intermediate, Old profiles);
- Mozilla, OpenSSH guidelines: https://infosec.mozilla.org/guidelines/openssh;
- CIS Benchmarks (give the benchmark name, without a recommendation number if you are not certain of it);
- official documentation of the software concerned.

If you have no reliable reference for a deviation, write "Reference: general good practice, to be sourced" rather than invent one. Never quote a control number you are not certain of.

## Output format

A deviation table, from most to least risky, with columns: location (`file:line`), directive, observed value, recommended value, risk (one sentence), reference, confidence (Confirmed by the file read, Probable, To verify).

Then:
- the compliant points, one line each, without detail;
- a "Not verified" section.

## Rules

- Never invent a directive, default value or line number. A default value depends on the version: if it is not visible, write "To verify".
- A missing directive is not necessarily a deviation: say what needs checking (default value for the version, included file).
- Adapt the recommendation to the stated compatibility profile; do not force a strict profile without stating its cost (clients excluded).
- Warn that any change must be tested before going to production (risk of cutting off access or clients).
- Mask secrets you come across: keep the first four characters and replace the rest with `…`.
- Write in the user's language, with no pleasantries or generic conclusion.

A configuration file is reviewed faster and more reliably with a checklist. This skill has the model compare the files you give it (SSH, nginx, Apache, TLS settings) with published recommendations, and report each deviation with its directive, its risk and the recommended value.

What the skill produces

  • A deviation table: file and line, directive, observed value, recommended value, risk, reference.
  • The points covered: SSH authentication and algorithms, TLS versions and cipher suites, security HTTP headers, information exposure (banners, default pages), HSTS, permissions and dangerous options.
  • A "Not verified" section: directives inherited from an included file you did not provide, live configuration that differs from the file.

To check the real state of an exposed service, complete the review with enumeration using Nmap within the authorised scope.

Install the skill

In Claude Code, for all your projects:

mkdir -p ~/.claude/skills/configuration-hardening
curl -fsSL -o ~/.claude/skills/configuration-hardening/SKILL.md \
  https://arsenal.bearops.fr/telechargements/skills/durcissement-configuration/en/SKILL.md

For a single project, put the file in .claude/skills/configuration-hardening/SKILL.md at the root of the repository. With another agent, add the file’s content to its project instructions.

Use it

Give the configuration and the usage context:

Check the hardening of these files (Debian 12, web server exposed to the Internet,
SSH access reserved for administration from a trusted network):
sshd_config and the nginx server block.

[paste the files or excerpts here, with line numbers if possible]

State the compatibility profile you are aiming for (old clients or not): the Mozilla recommendations distinguish several profiles, and the right choice depends on your users.

Limitations

  • The skill reads files: it sees neither the configuration actually loaded nor the clients actually connecting. Confirm a deviation with a check on the service (for example sshd -T or a TLS scan) before reporting it.
  • Recommendations change: check the current version of the guides cited before carrying a value into a report.
  • Applying a recommendation without testing can cut off legitimate clients: validate in pre-production.
  • Do not send any client configuration to an online AI service unless your contract allows it. Remove secrets, host names and internal addresses if needed.

Sources

Related resources

  • Methodology · Infrastructure and AD pentesting · Red teaming

    Active Directory config review

    Defensive Active Directory review checklist for audits: privileged accounts, krbtgt, delegation, LAPS, legacy protocols, GPOs, ADCS and logging.

    Published

  • Cheatsheet · Infrastructure and AD pentesting · Red teaming

    Active Directory enumeration

    Authenticated AD enumeration for an internal test: domain, accounts, groups, password policy, shares, LDAP queries and BloodHound collection.

    Published

  • AI prompt · Infrastructure and AD pentesting · OSINT · Cloud

    Attack surface prompt

    Prompt that analyses an asset inventory from the authorised scope (domains, IPs, services, cloud) to prioritise testing: exposure, criticality, checks.

    Published

BEAROPS

Need a penetration test?

BEAROPS experts test your applications, infrastructure and cloud, black box or grey box, following OWASP, NIST and ANSSI guidance.