Skip to content
  1. Home
  2. Agent skills
  3. security-audit

security-audit

Multi-phase security audit skill with independently verified findings

What it is

A skill from Cloudflare that runs a structured audit of one repository: reconnaissance of architecture and trust boundaries, coverage-tracked hunting by isolated sub-agents, validation of candidates, and independent verification before a machine-readable report is written. It is the single-repo starting point of Cloudflare's internal vulnerability discovery harness.

Good for

  • Audit your own repository for vulnerabilities
  • Map trust boundaries and input surfaces
  • Produce verified, machine-readable findings

Install or run

npx skills add https://github.com/cloudflare/security-audit-skill

Command as given in the project README. Read the repository before you run anything on your machine.

What you need

A coding agent that can spawn sub-agents; Node.js for npx

Worth knowingRuns many agent passes, so a full audit is slow and token-heavy; findings still need human triage.

Sources

The summary is ours, written from the README. Spotted an error? See how we work.

More for security & quality

See the topic →
  • verification-loop

    affaan-m/ECC

    Six-phase check of an agent's work ending in a PASS or FAIL report

    SkillGitHub stars: 302k
  • Requesting Code Review

    NousResearch/hermes-agent

    Pre-commit review with security scan, quality gates and auto-fix

    SkillGitHub stars: 263k
  • Web Pentest

    NousResearch/hermes-agent

    Authorized web pentest: recon, proof-based exploits and a report

    SkillGitHub stars: 263k
  • webapp-testing

    anthropics/skills

    Test local web apps with Playwright scripts

    SkillGitHub stars: 194k