Cloudflare publishes security-audit skill for coding agents
Cloudflare has published an MIT-licensed coding-agent skill called security-audit that, according to its repository README surfaced on GitHub Trending, turns a coding agent into a security auditor. The skill runs a six-phase workflow — reconnaissance, coverage-led hunting, candidate validation, structured output, independent record verification and target-neutral reporting — with isolated agents and adversarial validation in which the checker is never the finder. Findings are written to a schema-validated JSON file and checked by zero-dependency Node validators, and Cloudflare says the skill is the single-repo starting point from which its fleet-wide vulnerability discovery harness grew. Installation is via the Skills CLI, and running target code requires a model with tool use and parallel sub-agents plus an OS-enforced sandbox.
Cloudflare says the skill seeded its own multi-stage, fleet-wide vulnerability discovery harness, making it a concrete template for agent-driven security auditing.
A coding-agent skill for multi-phase security audits with independently verified, machine-readable findings
security-audit
A coding-agent skill that turns your agent into a security auditor. It orchestrates isolated agents through reconnaissance, coverage-led hunting, candidate validation, structured output, independent record verification, and target-neutral reporting.
This is the skill that seeded Cloudflare's vulnerability discovery harness, described in Build your own vulnerability harness. The harness grew into a multi-stage, fleet-wide system; this skill is the single-repo starting point it evolved from.
What it does
The skill runs a structured audit in six phases: