사용 방법
- 설치 명령을 복사하고 자신의 에이전트 환경에 설치하기 전에 소스 패키지를 확인하세요.
- 필요한 도구, 인증 정보, 지원 환경은 원본 문서에서 확인하세요.
- 입력과 원하는 산출물을 설명하고 출처와 결과를 대조하세요.
npx skills add https://github.com/smartworldarafath/rea_skill --skill website-security-audit시작 전 확인
- 설치에는 모델 구독, API 한도, 파일 접근 권한이 포함되지 않습니다.
- 별점 수와 설치 수는 활동 지표이며 독립적인 품질·보안 검증이 아닙니다.
원본 저장소의 제목, 예시, 코드, 표, 링크, 이미지를 유지해 표시합니다.
Website Security Audit & 10-Layer Web Infrastructure Scanner
Overview
This skill provides a standardized, automated methodology for performing defensive security audits, OWASP Top 10 vulnerability assessments, and full-stack architecture inspections on websites and web applications owned by the user. It combines Pentest-Tools Website Scanner DAST/SAST workflows (Light Passive Scan vs. Deep Active Scan, Spidering, Technology Fingerprinting, and Authenticated Testing) with our 5-Phase Defensive Audit and 10-Layer Web Stack Reference.
1. Scan Modes & Diagnostic Questions (Pentest-Tools Scanner Methodology)
When auditing a target URL or web repository, support three execution modes (defaulting to Deep Full-Stack Audit unless specified):
- Light Scan (Passive & Non-Intrusive):
- Inspects HTTP response headers, DNS/SSL/TLS posture, server banners, cookie security flags (
HttpOnly,Secure,SameSite), technology fingerprinting, and client-side JS bundle secret leaks without sending active fuzzing payloads.
- Deep Scan (Active OWASP Top 10 Assessment):
- Performs automated crawling/spidering to discover hidden routes, forms, API endpoints, and query parameters.
- Evaluates OWASP Top 10 risks: SQL Injection (SQLi), Cross-Site Scripting (Reflected/Stored/DOM XSS), OS Command Injection, Local/Remote File Inclusion (LFI/RFI), SSRF, Broken Access Control (IDOR/BOLA), and Outdated JS Libraries / CVE exposure.
- Authenticated Scan (Session / Cookie / Bearer Token):
- Audits protected dashboards, multi-tenant isolation (Supabase/Postgres RLS), state-changing POST/PUT/DELETE endpoints, and payment webhooks behind login boundaries.
Smart Diagnostic Questions (When User Asks "How Should I Audit My Website?")
If the user asks what to check or how to prompt, offer or automatically verify these core questions:
- Are HTTP security headers (`HSTS`, enforced `CSP`, `X-Frame-Options`, `X-Content-Type-Options`, strict `CORS`) properly blocking XSS and clickjacking?
- Are any private keys (`SUPABASE_SERVICE_ROLE_KEY`, `sk_live_
, database URLs, internal.env` variables) bundled into client-side Webpack/Vite/Next.js chunks?* - If `anon_key` or Firebase config is public, is Row-Level Security (RLS) strictly enforced on every table (`auth.uid() = user_id`)?
- Are public `/api/`, GraphQL, and authentication endpoints protected against brute-force, credential stuffing, and denial-of-wallet attacks via rate limiting?*
- Do Stripe/Paddle/LemonSqueezy webhook handlers verify cryptographic HMAC signatures before granting user entitlements?
2. Automated 5-Phase Defensive Audit Pipeline
Phase 1: HTTP Security Headers & Transport Security
Inspect response headers using curl.exe -I -s -L https://<target-domain>/ or REA MCP inspect_web_page:
- Strict-Transport-Security (HSTS): Must include
max-age=31536000; includeSubDomains; preload. - Content-Security-Policy (CSP):
- Flag if CSP is only in
Content-Security-Policy-Report-Only(non-blocking) instead of enforcedContent-Security-Policy. - Flag dangerous directives:
'unsafe-inline','unsafe-eval', or overly broad wildcards (*). Recommend cryptographic nonces (nonce-<base64>).
- Cross-Origin Resource Sharing (CORS): Flag
Access-Control-Allow-Origin: *or reflectedOriginwithAccess-Control-Allow-Credentials: trueon authenticated routes. - Clickjacking & MIME Defense: Verify
X-Frame-Options: DENY(or CSPframe-ancestors 'none') andX-Content-Type-Options: nosniff. - Referrer & Permissions: Verify
Referrer-Policy: strict-origin-when-cross-originand restrictivePermissions-Policy.
Phase 2: Frontend Bundle & Secret Leakage Inspection
Use REA MCP (analyze_web_bundle, recover_javascript_sources, export_web_scripts) or automated Python/Node inspection to scan all HTML and .js chunks:
import urllib.request, re
url = "https://<target-domain>/"
req = urllib.request.Request(url, headers={"User-Agent": "Mozilla/5.0"})
html = urllib.request.urlopen(req).read().decode("utf-8", errors="ignore")
high_risk_patterns = {
"Supabase Service Role Key": r"eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9\.[a-zA-Z0-9_\-]+\.[a-zA-Z0-9_\-]+",
"Private API Key / Secret": r"(?:sk_live_|secret_key|private_key|api_secret)\w+",
"Database URL with Credentials": r"postgres://\w+:[^@]+@[\w\.-]+",
"AWS Access Key": r"AKIA[0-9A-Z]{16}",
}
api_routes = set(re.findall(r'["\'](/api/[\w\-/]+)["\']', html))- Safe to expose:
NEXT_PUBLIC_*, Supabaseanon_key(only when RLS is enabled), Clerk publishable key (pk_live_*). - Critical Leaks: Supabase
service_role_key, Stripesk_live_*/ webhook secrets (whsec_*), database credentials,.env/.git/HEADexposure.
Phase 3: Auth & Database Access Controls (RLS Verification)
- Row Level Security (RLS): Confirm all Supabase/PostgreSQL tables have
ENABLE ROW LEVEL SECURITYand policies restricting access byauth.uid() = user_id. - Session & JWT Security: Verify short-lived access tokens,
HttpOnly; Secure; SameSite=Lax/Strictcookies, and server-side session revocation.
Phase 4: API Endpoint, GraphQL & Route Hardening
- Authentication & Authorization (BOLA/IDOR): Verify unauthenticated requests to private
/api/*routes return401/403and users cannot access other tenants' IDs. - Rate Limiting & Schema Validation: Ensure login, OTP, AI generation, and checkout endpoints enforce rate limits (Upstash Redis / Cloudflare WAF) and strict schema validation (Zod / Pydantic).
- GraphQL Introspection: Check if production GraphQL endpoints leave full schema introspection and unbounded query depth open.
Phase 5: Webhook & Payment Security
- Webhook Signature Verification: Verify HMAC signatures (
stripe.webhooks.constructEvent, Paddle signature verification) using the raw request body before fulfilling orders. - Idempotency: Verify duplicate event IDs are deduplicated in the database.
3. Complete 10-Layer Website Infrastructure & Tooling Matrix
When performing deep web architecture recovery, reconnaissance, or security scanning, leverage this 10-layer ecosystem:
- DNS, Domains & SSL/TLS Infrastructure:
- Diagnostics:
dig,nslookup,whois,traceroute,nmap,Shodan,Censys,DNSChecker,MXToolbox. - TLS/Crypto:
openssl s_client,Certbot/ Let's Encrypt, Cloudflare SSL/TLS cipher grading.
- Web Servers, Reverse Proxies & Load Balancers:
- Servers & Proxies: Nginx, Apache, Caddy, Microsoft IIS, OpenLiteSpeed, HAProxy, Envoy Proxy, Traefik, AWS ALB/ELB, Cloudflare CDN.
- Web Hosting, Cloud Infrastructures & PaaS:
- Cloud & PaaS: AWS, GCP, Azure, Vercel, Netlify, Render, Railway, Fly.io, Cloudflare Pages, DigitalOcean, Hetzner, cPanel/Plesk.
- Containers, Orchestration & CI/CD:
- Runtime & Pipelines: Docker, Podman, LXC, Kubernetes (K8s), Helm, GitHub Actions, GitLab CI/CD, Jenkins, ArgoCD.
- Databases, Caches & Vector Stores:
- SQL/NoSQL/Vector: PostgreSQL, MySQL/MariaDB, SQLite, MongoDB, DynamoDB, Firestore, Redis, Kafka, Elasticsearch, Pinecone, Milvus, Chroma.
- Frontend, Backend Frameworks & CMS:
- Frontend: React, Next.js, Vue, Nuxt, Svelte, Tailwind CSS.
- Backend: Node.js (Express, NestJS, Fastify), Python (Django, FastAPI, Flask), PHP (Laravel), Go (Gin, Fiber), Java (Spring Boot), Ruby on Rails.
- CMS: WordPress, Strapi, Ghost, Drupal, Shopify, Magento.
- API Architecture, Testing & Documentation:
- Protocols & Specs: REST, GraphQL, gRPC, WebSockets, Webhooks, SOAP, OpenAPI / Swagger, Postman, Insomnia, Bruno, Hoppscotch.
- Web Traffic Inspection & Network Debugging:
- Intercepting Proxies: Burp Suite, OWASP ZAP, Caido, Charles Proxy, Proxyman,
mitmproxy, HTTP Toolkit, Wireshark,tcpdump.
- Web Security Scanners, Fuzzers & WAFs:
- Discovery & Scanning:
Nuclei,ffuf,Gobuster,Feroxbuster,Dirsearch,SQLmap,Nikto,WPScan,Amass,Sublist3r. - WAFs: Cloudflare WAF, ModSecurity, AWS WAF.
- Client-Side Web Reverse Engineering & Scraping:
- Inspection & Automation: REA MCP (
analyze_web_bundle,inspect_web_page,observe_web_session), Chrome DevTools,de4js,Retire.js, Microsoft Playwright MCP, Puppeteer, Bright Data MCP, Hyperbrowser MCP,Wappalyzer.
4. Deliverable Format
Always structure the output report with:
- Executive Summary & Risk Score
- Critical & High Priority Findings (with exact proof/headers/lines)
- Medium & Low Priority Hardening
- Verified Strengths
- Copy-Paste Remediation Code Snippets (
next.config.js,middleware.ts, SQL RLS policies, Nginx/Cloudflare rules).