fingerprinting

Identify exact technology stack, frameworks, versions, and CMS on live hosts. Use when the user has a live host list and needs to know what's running before choosing vuln classes to test.

SKILL.md

GitHub

Tech Stack Fingerprinting

Knowing it's Spring Boot 2.5 → CVE-2022-22965 candidate. Knowing it's "Java" → useless.

When to invoke

Trigger phrases:

  • "what's running on X"
  • "fingerprint this host"
  • "tech stack of target"
  • "version of framework"

What we identify

Layer Signals Tools
Web server Nginx, Apache, IIS, Caddy httpx, nmap, headers
Backend lang Node, Python, Java, .NET, Go, PHP, Ruby headers, cookies, errors
Framework Express, Django, Spring, Rails, Laravel, FastAPI error pages, default routes
CMS WordPress, Drupal, Joomla, Ghost paths, scripts, meta
Frontend React, Vue, Angular, Svelte, Next.js JS files, bundle structure
API style REST, GraphQL, gRPC, SOAP endpoints, schemas
Cloud provider AWS, GCP, Azure, Cloudflare, Fastly DNS, headers, error pages
WAF Cloudflare, Akamai, Imperva, AWS WAF headers, behavior
Dependencies npm, JS libraries, CDN versions retire.js

Step-by-Step Workflow

1. Quick scan with httpx

# tech-detect uses Wappalyzer logic
echo "https://app.target.com" | httpx -tech-detect -title -web-server -status-code -json | jq .

# Bulk on alive hosts:
cat live.txt | httpx -tech-detect -web-server -json | tee tech-detected.jsonl

Output sample:

{
  "url": "https://app.target.com",
  "tech": ["React", "Next.js:13.4", "Cloudflare", "Webpack"],
  "webserver": "cloudflare",
  "title": "Target App"
}

2. Deep scan with WhatWeb

# WhatWeb has wider plugin DB than httpx
whatweb -v --color=never -a 3 "https://app.target.com"

# Aggression levels:
#  1 = passive (just GET /)
#  3 = standard (default)
#  4 = heavy (lots of requests)

3. Header forensics

# Manual but precise:
curl -sI "https://app.target.com" | grep -iE "server|x-powered-by|x-aspnet|x-runtime|via|cf-ray|x-amz-|x-azure-"

# Examples of what to look for:
# X-Powered-By: PHP/7.4.3
# X-AspNet-Version: 4.0.30319
# X-Runtime: 0.045821 → Ruby on Rails
# Server: nginx/1.18.0 (Ubuntu)
# Via: 1.1 vegur → Heroku
# X-Vercel-Id → Vercel hosting
# X-Amz-Cf-Id → AWS CloudFront

4. Cookie fingerprinting

curl -sI "https://app.target.com" | grep -i "set-cookie"

# Cookie names reveal stack:
# PHPSESSID → PHP
# JSESSIONID → Java
# .AspNet.ApplicationCookie → ASP.NET
# _csrf, _session → Rails
# laravel_session → Laravel
# connect.sid → Express
# session → Flask/generic
# next-auth.session-token → Next.js + NextAuth

5. Error page fingerprinting

# Common error-page paths to fingerprint stack
for path in "/foo/bar" "/api/random-uuid" "/404-this-doesnt-exist" "/.env" "/server-status"; do
    echo "=== $path ==="
    curl -s "https://app.target.com$path" | head -30
    echo
done

# Look for:
# - "Whitelabel Error Page" → Spring Boot
# - "Application Error" + heroku → Heroku app
# - "Cannot GET /foo" → Express
# - "DEBUG = True" Django stack trace → DEBUG mode (CRITICAL bug!)
# - "Internal Server Error" + Werkzeug → Flask

6. Frontend stack from JS bundle

# Get JS file URLs
curl -s "https://app.target.com" | grep -oE 'src="[^"]+\.js"' | sed 's/src="//;s/"$//'

# Check for framework fingerprints:
# react-dom.production.min.js → React
# vue.runtime.min.js → Vue
# @angular/core → Angular
# next/router → Next.js
# nuxt → Nuxt
# svelte → Svelte
# remix-run → Remix

# retire.js — find outdated JS libs
retire --jsrepo https://retirejs.github.io/retire.js/repository/jsrepo.json --js https://app.target.com

7. Nuclei tech-detect templates

# nuclei has hundreds of tech-detect templates
nuclei -u "https://app.target.com" -tags tech -silent

# Targeted: known versionable software
nuclei -u "https://app.target.com" -t http/technologies/ -silent

# Look for version disclosure
nuclei -u "https://app.target.com" -tags version -silent

8. Cloud / CDN detection

# DNS shows CNAME → CDN/provider
dig +short "app.target.com"

# Common CNAMEs:
# *.cloudfront.net → AWS CloudFront
# *.azureedge.net → Azure CDN
# *.fastly.net → Fastly
# *.cloudflare.com → Cloudflare (often via A record + IPs in CF range)
# *.ngrok.io → ngrok (developer tunnel — interesting!)
# *.herokuapp.com → Heroku
# *.vercel.app → Vercel
# *.netlify.app → Netlify
# *.s3.amazonaws.com → S3 bucket (check ownership = takeover candidate)
# *.github.io → GitHub Pages (takeover candidate)

9. WAF detection

# wafw00f
pip install wafw00f
wafw00f "https://app.target.com"

# Manual:
# - 403 with "challenge" page → Cloudflare
# - 403 + cf-ray header → Cloudflare
# - 403 + akamai-x-cache header → Akamai
# - 406 + X-Iinfo → Imperva
# - 403 + X-Sucuri-ID → Sucuri

Output template

target: app.target.com
fingerprint_date: 2026-06-02

infrastructure:
  cdn: Cloudflare
  ip: 104.21.x.x (CF IP)
  origin_ip: unknown (CF protected)
  waf: Cloudflare WAF

web_server:
  product: nginx
  version: 1.18 (inferred from defaults)
  os: Ubuntu 20.04 LTS

backend:
  language: Node.js
  framework: Next.js 13.4
  evidence: |
    - X-Powered-By: Next.js
    - JS bundle: _next/static/
    - Headers: x-vercel-id (suggests Vercel hosting backed by Next.js)

frontend:
  framework: React 18.2
  bundler: Webpack 5
  state_mgmt: Redux Toolkit (from JS analysis)

api:
  style: REST + GraphQL
  rest_base: /api/v3
  graphql_endpoint: /api/graphql
  evidence: discovered in main.js bundle

authentication:
  type: NextAuth.js with custom OAuth provider
  cookie: next-auth.session-token
  jwt: yes (verified via JWT decoding)

cms: none

interesting_tech:
  - GraphQL endpoint exposed (introspection?)
  - NextAuth handles session — check JWT signing
  - Vercel serverless — different attack surface (cold starts, env vars)
  - WebSocket on /api/ws — possible RT data exchange

Tech → vuln class lookup

Use this to decide what to hunt next:

Tech Found Try These Skills
WordPress nuclei wordpress templates, wpscan, REST API authcheck
GraphQL [[graphql]] — introspection, batching, depth
JWT in cookies [[jwt-attacks]] — alg=none, key confusion
OAuth visible [[oauth-oidc]] — redirect_uri, state
File upload UI [[file-upload]]
Multi-tenant URLs [[idor-hunting]]
Search/filter inputs [[sqli]], [[ssti]]
URL fetcher / webhook [[ssrf]]
Spring Boot CVE checks (RCE history), actuator endpoints
Old Drupal CVE-2018-7600 Drupalgeddon, etc.
Old WordPress plugins nuclei wordpress-CVEs templates
Outdated jQuery retire.js → CVEs
Exposed Swagger/OpenAPI API enumeration goldmine

Cross-references

  • [[asset-discovery]] — runs before this
  • [[js-analysis]] — deeper JS endpoint mining
  • [[content-discovery]] — fuzz paths based on tech
  • [[nuclei]]-style scanning — match templates to fingerprinted stack

Common pitfalls

  1. Trusting Server: header alone. It's often reverse-proxy, not the actual app.
  2. Missing JS-based SPA detection. SPA = different attack surface.
  3. Not checking all origin endpoints. Same domain may have multiple stacks behind path-based routing.
  4. Treating Cloudflare as the app. WAF can mislead — try origin IP if findable.
  5. Skipping the error pages. Stack traces > Wappalyzer fingerprints any day.

Pro tips

  • Bring up devtools in the browser. Network tab shows headers + cookies in seconds.
  • Note _next/, __nuxt/, _remix/ paths. They signal SSR frameworks → unique bug classes.
  • Save fingerprint to loot/<target>/fingerprint.yaml. Re-use for re-tests.
  • Run fingerprinting weekly on active targets. Tech stack updates create new attack windows.