security-audit

An agent skill by cloudflare, from cloudflare/security-audit-skill. Tags: api, cli, debugging, devops, research, security.

What it does

Security guidance and vulnerability review for codebases, APIs, services, CLI tools, libraries, and daemons. Use for security questions, focused reviews, vulnerability research, security audits, or pen tests. Run the complete workflow only for explicit codebase audit or pen-test requests, full/comprehensive/end-to-end reviews, or requested report artifacts.

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add cloudflare/security-audit-skill --skill security-audit

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/cloudflare/security-audit-skill
cp -r security-audit-skill/skills/security-audit ~/.claude/skills/security-audit

Safety box score

Grade C, moderate risk: 33 of 100 risk points, rated 2026-10-19. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This skill is a multi-phase security-audit orchestrator: it can run the target’s builds/tests and other tools in a sandbox, then write retained run artifacts and reports. The main risks are local code execution on untrusted project code, local filesystem footprint outside the repo, and incomplete visibility into the shipped validation scripts.

Category levels, from 0 (none) to 3 (high)
Category	Level	Why
Code & shell execution	2 (moderate)	It explicitly runs target-controlled builds, tests, processes, browsers, emulators, fuzzers, and fixture processing inside a sandbox, so installing it can lead to controlled execution of repository code.
Network access	1 (low)	The workflow allows only isolated loopback traffic and forbids external network access, so network reach is narrow and local-only.
Filesystem reach	2 (moderate)	It writes run metadata, ledger, reports, and per-agent scratch/artifact trees, with the default output directory under `~/security-audit-skill/<repo-name>/run-<N>` outside the target repo.
Secrets & credentials	0 (none)	The skill uses dummy secrets and forbids probing live credentials, keychains, or external identity systems; no direct secret-store access is requested.
Privilege & persistence	0 (none)	It does not request sudo, service installation, or shell/OS setting changes; dependency installs and system-wide mutation are explicitly disallowed.
Instruction hijack surface	1 (low)	The files are prompt-heavy, but they do not instruct the agent to ignore safeguards or fetch remote instructions; the main concern is normal agent-guidance text, not a clear injection payload.
Irreversible actions	1 (low)	It can autonomously run a full six-phase audit and generate retained artifacts, but it does not itself push, deploy, message, or spend external resources.
Transparency	2 (moderate)	Four JS files were not inspected (`validate-coverage-ledger.cjs`, `validate-coverage-ledger.test.cjs`, `validate-findings.cjs`, `validate-findings.test.cjs`), leaving the enforcement code for validation opaque.

Before installing

Read `validate-findings.cjs` and `validate-coverage-ledger.cjs` before trusting the skill’s report validation.
Expect run artifacts under `~/security-audit-skill/<repo-name>/run-<N>` and confirm that local writes there are acceptable.
If you will use full audit mode, try it first on a disposable repo copy because it executes the target’s code in a sandbox.

Rated 2026-10-19 by an automated review of 16 files from the skill's repository.

Source

Repository: cloudflare/security-audit-skill (all skills from this repository)
Path: skills/security-audit/SKILL.md
Branch: main
Updated: 2026-09-19

address-sanitizer — Builds and runs code under AddressSanitizer to catch buffer overflows, use-after-free, and other memory errors during fuzzing or tests.
modern-cpp — Guides C++ code toward modern idioms (C++20/23/26). Use when writing new C++ code, modernizing legacy patterns, or working on security-critical C++.
chaos-engineer — Designs chaos experiments, creates failure injection frameworks, and facilitates game day exercises for distributed systems — producing runbooks.
code-review — Review code for actionable defects. Correctness is the core; performance and security are optional sub-cases of the same engine.
java-architect — Use when building, configuring, or debugging enterprise Java applications with Spring Boot 3.x, microservices.
kubernetes-specialist — Use when deploying or managing Kubernetes workloads.