High signalMITCommunity

Security Audit

Senior security auditor subagent for comprehensive security assessments, compliance audits, and risk evaluations across systems, infrastructure, and processes.

Entry verified April 21, 2026

The short answer

Runs systematic vulnerability analysis, compliance-gap identification, and evidence-based security findings. Covers SOC 2, ISO 27001/27002, HIPAA, PCI DSS, GDPR, NIST, and CIS benchmarks with a defined audit scope, thorough controls assessment, and actionable remediation recommendations.

When to use it

Invoke when you need a vulnerability assessment (network scanning, application testing, configuration review, patch management), an access-control audit (privilege analysis, segregation of duties, MFA, password policies), a data-security review (classification, encryption, retention, DLP), or an incident-response readiness check.

Setup

  1. 1

    Install from VoltAgent/awesome-claude-code-subagents — the security-auditor lives in categories/04-quality-security/.

  2. 2

    Ships with model: opus and a restricted tool set (Read, Grep, Glob) so it audits without modifying the repo.

  3. 3

    Expects a context manager to supply security policies and compliance requirements at invocation.

  4. 4

    Output is a structured audit report: scope, controls assessed, vulnerabilities, compliance findings, and prioritised remediation steps with evidence.

Example

You: /security-review
Claude: 3 findings.
       HIGH: .env.example contains a real-looking AWS key (leak risk if copied).
       MED: Contact form /api/contact lacks rate limiting.
       LOW: Next.js headers config missing strict-transport-security.

Source & attribution

Author
VoltAgent
Licence
MIT
Type
Community

Reused under a permissive licence. Preserve attribution when forking.

Caveats

Not a pen-test. Rule-based scanner with LLM judgement — can miss logic-level exploits.

New skills, explained plainly.

One short email when a new Claude Code skill is worth installing — and when to skip the hype. No spam.

56 skills across 10 categories, all licence-verified.