Run #11 Report | GuardDex TestBench

75passed 1failed 9skipped 94total 79% pass rate

⚠ Vulnerabilities Found (3)

LOW Potential XSS via URL reflection

Script tag in URL may be reflected

Fix: Ensure JSON responses have correct Content-Type

MEDIUM Open Redirect Abuse

Trusted domain open redirects to malicious pages not flagged

Fix: Parse redirect destination URL and scan that too

HIGH Evasion Technique Bypass

AI-generated evasion URLs bypassing GuardDex rule engine

Payload: https://paypa1.attacker.com/login.php https://appl3.attacker.domain/account?action=signin https://amaz0n.attacker.com/order-confirmation https://paypa1%2eattacker.com/login.php https://appl3.attacker%2edomain/account?action=signin

Fix: These bypassed URLs reveal specific gaps — add to regression tests

Smoke ( 6 pass / 0 fail / 6 total)

✔ health_endpoint API healthy | 44 brands | version 2.0.0 4ms

✔ root_endpoint API root OK: GuardDex API v2.0.0 2ms

✔ docs_accessible OpenAPI docs accessible at /docs 1ms

✔ brands_endpoint 44 protected brands loaded 3ms

✔ challenge_stats Stats OK: 376 challenges issued 3ms

✔ dashboard_accessible Challenge screen HTML reachable 3ms

Phishing ( 12 pass / 0 fail / 12 total)

✔ fake_paypal CAUGHT → block | risk=critical | 3 signals | Would have saved David $200 4ms

✔ fake_facebook CAUGHT → block | Would have stopped the 2-year hack chain 3ms

✔ real_paypal Real PayPal correctly ALLOWED 3ms

✔ ip_login IP address login BLOCKED | signals: ['ip_address_login', 'brand_impersonation', 'http_login_page'] 4ms

✔ punycode Punycode domain BLOCKED 2ms

✔ official_sites_allow All 10 official sites correctly ALLOWED 1056ms

✔ phishing_bulk All 24 phishing URLs caught (100% detection rate) 2489ms

✔ 37_brands_stress All 37/37 brands detected (100%) 72718ms

✔ long_subdomain_evasion Deep subdomain attack caught: block 3ms

✔ double_extension Double extension caught: block 2ms

✔ https_not_safe HTTPS phishing site still blocked - good 2ms

✔ signal_quality 3 well-formed signals returned: ['suspicious_tld', 'brand_impersonation', 'http_login_page'] 2ms

Challenge ( 13 pass / 0 fail / 13 total)

✔ correct_domain Correct domain accepted | official_url=https://www.paypal.com 2ms

✔ alternate_domain https://paypal.com (no www) accepted 1ms

✔ fake_domain Fake domain correctly rejected: That is not an official login address for PayPal 1ms

✔ http_rejected HTTP (non-HTTPS) correctly rejected by challenge 1ms

✔ ip_rejected IP address correctly rejected by challenge 1ms

✔ punycode_rejected Punycode domain rejected by challenge 2ms

✔ empty_input Empty input rejected 2ms

✔ wrong_brand Cross-brand domain correctly rejected 24ms

✔ subdomain_official Official domain with path accepted 2ms

✔ xss_input XSS input rejected by challenge validator 2ms

✔ sqli_brand SQL injection in brand field rejected (HTTP 400) 2ms

✔ google_challenge Google challenge: correct domain accepted 2ms

✔ apple_challenge Apple ID challenge: correct domain accepted 2ms

Security ( 11 pass / 0 fail / 13 total)

✔ cors_wildcard CORS restricted to: https://evil.example.com 1ms

✔ server_header Server header: 'guarddex' (acceptable) 3ms

✔ 404_no_trace 404 response clean (HTTP 404) 2ms

✔ malformed_json Malformed JSON handled gracefully (HTTP 422) 2ms

✔ empty_body Empty body correctly rejected (422/400) 4ms

✔ long_url Very long URL rejected (HTTP 400) 2ms

✔ null_bytes Null byte URL rejected (HTTP 400) 2ms

✔ js_protocol javascript: URI rejected at HTTP level (400) 1ms

✔ data_uri data: URI rejected at HTTP level (400) 2ms

⚠ rate_limit No rate limiting detected on scan endpoint (0/25 throttled) - OK for internal API 55ms

✔ scan_get_rejected GET on POST-only endpoint returns 405 2ms

✔ brands_post_rejected POST on GET-only brands endpoint returns 405 1ms

⚠ response_xss URL with script tag reflected in response (check content-type header) 3ms

Ai ( 1 pass / 0 fail / 10 total)

ℹ v2_detection v2.0 endpoints not yet on target (expected - they're being merged). Remaining AI tests will run in probe/compatibility mode. 5ms

✔ ollama_reachable AI analysis field in response (4704ms) - Ollama connected 4703ms

– v2_register v2.0 not yet on target 5ms

– weak_password v2.0 not yet on target 6ms

– dupe_email v2.0 not yet on target 7ms

– wrong_password v2.0 not yet on target 5ms

– jwt_manipulation v2.0 not yet on target 4ms

– ai_typosquatting v2.0 not yet on target 5ms

– scan_performance v2.0 not yet on target 7ms

– user_enum v2.0 not yet on target 5ms

Analytics ( 11 pass / 0 fail / 12 total)

✔ dashboard_reachable Analytics dashboard OK (HTTP 200) 3ms

✔ dashboard_structure All 5 required keys present 3ms

✔ dashboard_stats_fields Stats valid — 44 brands protected, 1648 total scans 2ms

✔ dashboard_trend 30-day trend: 30 labels, correct MM-DD format 3ms

✔ stats_consistency Stats consistent: 1648 total = 781 blocked + 414 challenged + 453 allowed 3ms

✔ ai_status_endpoint Ollama available — model=phi3:mini loaded=['phi3:mini'] 14ms

✔ ai_fields_present ai_used=False ai_analysis=null (expected for safe) 3ms

✔ ai_triggers_challenge AI ran in 4781ms — verdict=phishing confidence=90% 4780ms

✔ ai_skips_safe AI correctly skipped for official PayPal (action=allow, ai_used=False) 3ms

– ai_verdict_structure AI did not run (Ollama may be down) 2ms

✔ stats_increment total_scans incremented: 1655 → 1656 5ms

✔ challenge_stats Challenge stats OK — 417 challenged, 782 blocked, 1656 total 2ms

Browser_exploits ( 5 pass / 0 fail / 6 total)

✔ fake_browser_updates All 10 BeEF/fake-update delivery URLs caught (100%) 34390ms

⚠ drive_by_downloads Caught 5/7 (71%) — missed: fake Flash update → allow; fake Java install → allow 19849ms

✔ watering_hole All 4 watering hole URLs caught 20367ms

✔ browser_allow All 3 legitimate browser sites correctly allowed 7ms

✔ beef_running BeEF is running on :3000 — live hook tests available 19ms

✔ beef_hook_delivery All 4 BeEF hook delivery URL patterns caught 21896ms

Social_engineering ( 5 pass / 0 fail / 7 total)

✔ credential_harvesters All 9 credential harvester URLs caught (100%) 12143ms

✔ tech_support_scams All 7 tech support scam URLs caught (100%) 15298ms

✔ gov_impersonation All 6 government impersonation URLs caught (100%) 5135ms

✔ package_pretext All 6 package/prize pretext URLs caught 26468ms

⚠ romance_scams Caught 2/3 (67%) romance scam URLs — difficult without AI analysis of content 5014ms

✔ gov_allow All 4 legitimate government/service sites correctly allowed 9ms

ℹ set_available SET not installed (optional). Install: sudo apt install set or: git clone https://github.com/trustedsec/social-engineer-toolkit 0ms

Evasion ( 5 pass / 0 fail / 8 total)

✔ encoding_bypass All 4 URL-encoded evasions caught 16429ms

✔ subdomain_stacking All 6 subdomain stacking attacks caught (100%) 16ms

✔ homoglyph All 5 homoglyph attacks caught 12846ms

⚠ open_redirect Open redirect handling: 1 blocked, 0 challenged, 3 allowed — open redirects on trusted domains are hard to detect 12ms

⚠ shortener_challenge 3/4 URL shorteners allowed through without challenge — destination is unknown, risk cannot be assessed 10ms

⚠ typosquatting Caught 6/8 (75%) — missed: google triple-o; facebook 00→oo 22027ms

✔ combo_attacks All 3 multi-technique evasion attacks caught 4329ms

✔ evasion_fp_check All 3 legitimate sites still correctly allowed after evasion tests 7ms

Ai_redteam ( 6 pass / 1 fail / 7 total)

✔ attacker_model Attacker model: dolphin-mistral:7b — adversarial generation ready 20ms

✔ redteam_banks Model: dolphin-mistral:7b | Generated 7 URLs in 19683ms | GuardDex caught 7/7 (100%) 32167ms

✔ redteam_elderly dolphin-mistral:7b | 8 elderly-targeted URLs | 8/8 caught (100%) 32782ms

✔ gap_analysis No bypass gaps found in test set — all 4 caught by GuardDex 17708ms

✔ defender_ai_supplement phi3:mini caught 4 ambiguous scam URLs that needed AI context 24059ms

✔ stress_volume 15 AI-generated URLs | 15 caught (100%) | avg 3355ms/scan | total 50323ms 58020ms

Run #11 — Full Report

⚠ Vulnerabilities Found (3)