The OWASP Top 10 is the industry's shared vocabulary for web risk. But a list is not a methodology — "Broken Access Control" is a category, not a to-do. This is my working translation of each item into things you actually do during an assessment.
1. Broken Access Control
The most frequent finding in real assessments, and the most under-tested.
Testing steps:
1. Map every route and its intended role (guest / user / admin)
2. Log in as user A. Try every route as user A
3. Repeat for user B on user A's resources (horizontal)
4. Try admin routes while logged in as user A (vertical)
5. Change the ID/object parameter: /profile/42 → /profile/43 (IDOR)
6. Try HTTP method override: PUT/PATCH/DELETE on GET-only routes
7. Change the role in the JWT/localStorage and replay
[!TIP]
IDOR (insecure direct object references) is the bread and butter of access control testing. Keep a list of numeric IDs you see in the UI — every single one gets incremented and decremented.
The fix is boring and correct: deny by default, check authorization server-side on every request, and use indirect reference maps where possible.
2. Cryptographic Failures
- TLS: check for weak ciphers, expired certs, HTTP→HTTPS redirects on all routes (not just the homepage)
- Secrets in frontend: API keys in JS bundles, JWTs signed with
none or leaked secrets
- Sensitive data at rest: passwords hashed with bcrypt/argon2? Session tokens in logs?
- Security headers:
Strict-Transport-Security, Content-Security-Policy, X-Content-Type-Options
curl -I http://target/ | grep -i location
nmap --script ssl-enum-ciphers -p 443 target
3. Injection
Beyond SQL (see my SQL injection deep-dive):
- Command injection: inputs that reach
exec, shell, ping wrappers. Test ; ls, | id, $(whoami) in time-based style.
- LDAP / NoSQL:
' OR '1'='1, {"$ne": null} on login forms
- Template injection:
{{7*7}}, ${7*7} on error pages, email personalization, 404s
The common thread: any string you control that leaves the language layer is an injection candidate.
4. Insecure Design
This is the one you can't fix with a scanner. It's about the design being unsafe:
- Password reset flows that accept the email in a way that leaks account existence
- No rate limiting on auth endpoints (brute-force by design)
- Business logic: coupon codes that can be applied twice, negative quantities
My test: write the abuse case before you write the exploit. If you can describe a legitimate-sounding flow that loses money or data, the design is broken.
5. Security Misconfiguration
- Default credentials (admin/admin, pi/raspberry)
- Verbose errors (stack traces = a map of your app)
- Unnecessary endpoints: /actuator, /debug, /phpinfo.php, /admin
- CORS: Access-Control-Allow-Origin: * with credentials
- Directory listing on /uploads
- .git, .env, .DS_Store, *.bak on the web root
for f in .git .env robots.txt .DS_Store wp-login.php actuator; do
echo -n "/$f → "
curl -sk -o /dev/null -w "%{http_code}" https://target/$f; echo
done
6. Vulnerable & Outdated Components
nmap -sV -p- target
trivy fs . --severity CRITICAL,HIGH
The finding to report is never "old version" — it's "old version with known CVE-2024-XXXX, exploitable from here."
7. Identification & Authentication Failures
- Rate limiting on login, signup, password reset (try 50 rapid attempts)
- Password policy: minimum length, breach-list checks
- MFA: is it bypassable by editing the request? Can you downgrade to "email only"?
- Account lockout: does it exist, and is it per-account or per-IP?
8. Software Integrity Failures
- CI/CD artifacts: are release builds signed? Can you tamper a published bundle?
- Deserialization: Java
ObjectInputStream, Python pickle, PHP unserialize on user input
- Automated patching: does the vendor flow actually ship?
9. Insufficient Logging & Monitoring
Test the response, not just the presence:
1. Perform a flagged action (500 errors, 20 failed logins)
2. Check if it was detected: rate limit, alert, account flag
3. Check if it's logged: does the next incident have an audit trail?
[!NOTE]
An assessment is not complete until you can answer: "When the bad thing happened, did the team know within a reasonable time?" If no — that is a finding.
10. Server-Side Request Forgery (SSRF)
Any parameter that fetches a URL for you:
https://target/fetch?url=http://169.254.169.254/latest/meta-data/ (AWS metadata)
https://target/fetch?url=http://127.0.0.1:6379/ (internal Redis)
https://target/webhook?url=file:///etc/passwd
Techniques when filters exist: DNS rebinding, decimal IP (http://2130706433/), http://[::1], @ tricks. The goal is never "see /etc/passwd" — it's reach the internal network.
My Assessment Flow
Scope → passive recon → map endpoints → authz matrix
→ injection sweep → misconfig sweep → business logic
→ report (risk, evidence, fix) → retest
Every finding in a report needs: what, where, why it matters, how to fix. A scanner's output is a lead, never a finding.
Discussion (0)
Sign in or create an account to join the discussion.
No comments yet. Start the discussion.