Scanner deep-dive
Qwen 3.6 35B by Alibaba Qwen ↗
General-Purpose LLM · agentic-v1 · scored on 65/66 repositories. Strict scoring (unfinished repos counted as misses).
Methodology note
Locally hosted. Open-weight GGUF (Q4_K_M) via Ollama on one RTX PRO 6000, standard agentic-v1 harness. Cost $0 (self-hosted).37.0
F3 (strict)
38.7
F2 (strict)
35.5%
Recall (strict)
60.8%
Precision
65/66
Repos scored
qwen3.6-35b
Model
Free
Total cost
222s
Avg latency
§
Per-repository breakdown
Each bar shows true positives, false positives, and misses on one repository; bar length is proportional to that repo's labeled vulnerabilities. Ranked by F2.
| Repository | TP | FP | FN | Recall % | F2 |
|---|---|---|---|---|---|
| pythonssti | 2 | 1 | 0 | 100.0 | 90.9 |
| vfapi | 8 | 2 | 1 | 88.9 | 87.0 |
| dsvw | 19 | 4 | 8 | 70.4 | 72.5 |
| vulnerable-api | 10 | 4 | 4 | 71.4 | 71.4 |
| intentionally-vulnerable-python-application | 5 | 3 | 2 | 71.4 | 69.4 |
| insecure-web | 6 | 3 | 3 | 66.7 | 66.7 |
| vulnpy | 49 | 6 | 31 | 61.3 | 65.3 |
| python-insecure-app | 5 | 2 | 3 | 62.5 | 64.1 |
| vampi | 10 | 8 | 5 | 66.7 | 64.1 |
| vulnerable-flask-app | 13 | 8 | 8 | 61.9 | 61.9 |
| vc-kimi-code-seeded-v2-legal-case-django | 14 | 2 | 12 | 53.8 | 58.3 |
| vc-codex-high-seeded-v2-crm-saas-django | 14 | 7 | 11 | 56.0 | 57.9 |
| owasp-web-playground | 15 | 5 | 13 | 53.6 | 56.8 |
| vulnerable-python-apps | 12 | 8 | 10 | 54.5 | 55.6 |
| python-app | 11 | 5 | 10 | 52.4 | 55.0 |
| vc-codex-high-seeded-v2-legal-case-django | 13 | 8 | 12 | 52.0 | 53.7 |
| dsvpwa | 16 | 7 | 16 | 50.0 | 53.0 |
| vc-kimi-code-seeded-v2-support-desk-fastapi | 14 | 6 | 14 | 50.0 | 53.0 |
| vc-claude-code-seeded-v2-crm-saas-django | 14 | 7 | 14 | 50.0 | 52.6 |
| vulnerable-tornado-app | 7 | 5 | 7 | 50.0 | 51.5 |
| vc-codex-seeded-v2-logistics-dispatch-fastapi | 15 | 11 | 15 | 50.0 | 51.4 |
| vc-claude-code-seeded-v2-education-lms-django | 15 | 5 | 17 | 46.9 | 50.7 |
| vc-kimi-code-seeded-v2-property-management-fastapi | 14 | 6 | 16 | 46.7 | 50.0 |
| threatbyte | 12 | 7 | 14 | 46.2 | 48.8 |
| vc-kimi-code-seeded-v2-fintech-lending-fastapi | 15 | 7 | 18 | 45.5 | 48.7 |
| vc-kimi-code-seeded-v2-education-lms-django | 13 | 9 | 15 | 46.4 | 48.5 |
| dvblab | 10 | 7 | 12 | 45.5 | 47.6 |
| vc-kimi-code-seeded-v2-crm-saas-django | 12 | 6 | 15 | 44.4 | 47.6 |
| vc-kimi-code-seeded-v2-marketplace-commerce-fastapi | 12 | 6 | 15 | 44.4 | 47.6 |
| vc-codex-high-seeded-v2-hr-payroll-django | 11 | 5 | 14 | 44.0 | 47.4 |
| vc-claude-code-seeded-v2-logistics-dispatch-fastapi | 15 | 13 | 18 | 45.5 | 46.9 |
| vc-codex-high-seeded-v2-support-desk-fastapi | 12 | 6 | 16 | 42.9 | 46.2 |
| vc-codex-seeded-v2-hr-payroll-django | 16 | 2 | 23 | 41.0 | 46.0 |
| vc-claude-code-seeded-v2-legal-case-django | 13 | 7 | 18 | 41.9 | 45.1 |
| vc-codex-high-seeded-v2-fintech-lending-fastapi | 13 | 15 | 16 | 44.8 | 45.1 |
| vc-claude-code-seeded-v2-healthcare-clinic-django | 12 | 6 | 17 | 41.4 | 44.8 |
| vc-codex-high-seeded-v2-logistics-dispatch-fastapi | 12 | 8 | 17 | 41.4 | 44.1 |
| vc-claude-code-seeded-v2-hr-payroll-django | 11 | 7 | 16 | 40.7 | 43.7 |
| lets-be-bad-guys | 10 | 9 | 14 | 41.7 | 43.5 |
| vc-kimi-code-seeded-v2-hr-payroll-django | 12 | 10 | 17 | 41.4 | 43.5 |
| vc-codex-seeded-v2-fintech-lending-fastapi | 15 | 10 | 22 | 40.5 | 43.4 |
| vc-codex-seeded-v2-legal-case-django | 13 | 7 | 20 | 39.4 | 42.8 |
| vc-codex-high-seeded-v2-education-lms-django | 10 | 11 | 15 | 40.0 | 41.3 |
| vc-kimi-code-seeded-v2-healthcare-clinic-django | 12 | 14 | 18 | 40.0 | 41.1 |
| vc-codex-seeded-v2-education-lms-django | 13 | 10 | 22 | 37.1 | 39.9 |
| extremely-vulnerable-flask-app | 11 | 1 | 21 | 34.4 | 39.3 |
| vc-codex-seeded-v2-property-management-fastapi | 11 | 8 | 20 | 35.5 | 38.5 |
| vc-claude-code-seeded-v2-fintech-lending-fastapi | 10 | 5 | 19 | 34.5 | 38.2 |
| vc-claude-code-seeded-v2-support-desk-fastapi | 11 | 4 | 23 | 32.4 | 36.4 |
| vc-codex-seeded-v2-marketplace-commerce-fastapi | 10 | 18 | 19 | 34.5 | 34.7 |
| damn-vulnerable-flask-application | 5 | 8 | 10 | 33.3 | 34.2 |
| flask-xss | 9 | 3 | 21 | 30.0 | 34.1 |
| dvpwa | 6 | 7 | 17 | 26.1 | 28.6 |
| djangoat | 11 | 11 | 41 | 21.1 | 23.9 |
| vulpy | 11 | 7 | 46 | 19.3 | 22.4 |
| vc-claude-code-seeded-v2-property-management-fastapi | 5 | 0 | 28 | 15.2 | 18.2 |
| pygoat | 0 | 0 | 78 | 0.0 | 0.0 |
| vc-claude-code-seeded-v2-marketplace-commerce-fastapi | 0 | 10 | 32 | 0.0 | 0.0 |
| vc-codex-high-seeded-v2-healthcare-clinic-django | 0 | 0 | 26 | 0.0 | 0.0 |
| vc-codex-high-seeded-v2-marketplace-commerce-fastapi | 0 | 14 | 25 | 0.0 | 0.0 |
| vc-codex-high-seeded-v2-property-management-fastapi | 0 | 0 | 26 | 0.0 | 0.0 |
| vc-codex-seeded-v2-crm-saas-django | 0 | 0 | 34 | 0.0 | 0.0 |
| vc-codex-seeded-v2-healthcare-clinic-django | 0 | 17 | 41 | 0.0 | 0.0 |
| vc-codex-seeded-v2-support-desk-fastapi | 0 | 18 | 30 | 0.0 | 0.0 |
| vc-kimi-code-seeded-v2-logistics-dispatch-fastapi | 0 | 0 | 31 | 0.0 | 0.0 |
§
Detection by severity
| Severity | TP | FP | FN | Recall % |
|---|---|---|---|---|
| Critical | 103 | 0 | 48 | 68.2 |
| High | 292 | 8 | 446 | 39.6 |
| Medium | 259 | 6 | 634 | 29.0 |
| Low | 21 | 1 | 64 | 24.7 |
§
Detection by vulnerability class
| CWE family | TP | FP | FN | Recall % |
|---|---|---|---|---|
| HTTP Header Injection | 2 | 0 | 0 | 100.0 |
| XPath Injection | 4 | 0 | 0 | 100.0 |
| Insecure Deserialization | 32 | 0 | 8 | 80.0 |
| Open Redirect | 31 | 0 | 9 | 77.5 |
| XML External Entities | 29 | 1 | 9 | 76.3 |
| Command / OS Injection | 34 | 0 | 13 | 72.3 |
| SQL Injection | 55 | 4 | 22 | 71.4 |
| Code Injection / RFI | 23 | 0 | 11 | 67.6 |
| Server-Side Request Forgery | 27 | 2 | 13 | 67.5 |
| Path Traversal | 29 | 3 | 14 | 67.4 |
| Cross-Site Scripting | 54 | 0 | 54 | 50.0 |
| Hardcoded Credentials | 31 | 0 | 36 | 46.3 |
| Security Misconfiguration | 46 | 0 | 65 | 41.4 |
| Other | 223 | 1 | 643 | 25.8 |
| Sensitive Data Exposure | 33 | 1 | 116 | 22.1 |
| Broken Access Control / IDOR | 11 | 0 | 70 | 13.6 |
| Denial of Service | 4 | 0 | 39 | 9.3 |
| Missing Authentication / Authorization | 7 | 3 | 70 | 9.1 |
§
LLM operational metrics
189,528
Avg input tokens
9,021
Avg output tokens
198,550
Avg total tokens
222s
Avg latency / repo
18.2%
JSON repair rate
66
Total runs
§
Cost
Free
Total cost
131,775
Python LOC scanned
65
Successful runs