{"slug": "dognavy-cybergym-technical-report", "title": "DoGNAVY CyberGym Technical Report", "summary": "DoGNAVY, a joint AI security harness by deepsec@DARKNAVY and independent security researchers based in Shanghai, passed 1,369 of 1,507 tasks (90.84%) in a full CyberGym Level 1 evaluation. The agentic system combines reachability analysis, proof-of-concept construction, dynamic testing, and independent review in a multi-agent workflow for vulnerability reproduction.", "body_md": "DoGNAVY is a joint AI security harness by [deepsec@DARKNAVY](https://deepsec.darknavy.net) and independent security researchers based in Shanghai. In a full CyberGym Level 1 evaluation, it passed 1,369 of 1,507 tasks (90.84%).\n\nThis report presents DoGNAVY, an agentic system for vulnerability reproduction, and documents its evaluation on the complete [CyberGym](https://www.cybergym.io/cybergym/) Level 1 task set. DoGNAVY combines reachability analysis, proof-of-concept (PoC) construction, dynamic testing, and independent review in a multi-agent workflow. The system can revise its approach as evidence accumulates, subject to explicit task boundaries and runtime controls.", "url": "https://wpnews.pro/news/dognavy-cybergym-technical-report", "canonical_source": "http://www.darknavy.org/blog/dognavy_cybergym_technical_report/", "published_at": "2026-08-05 06:57:28+00:00", "updated_at": "2026-08-05 08:23:24.431641+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-agents", "ai-safety", "ai-research"], "entities": ["DoGNAVY", "DARKNAVY", "CyberGym"], "alternates": {"html": "https://wpnews.pro/news/dognavy-cybergym-technical-report", "markdown": "https://wpnews.pro/news/dognavy-cybergym-technical-report.md", "text": "https://wpnews.pro/news/dognavy-cybergym-technical-report.txt", "jsonld": "https://wpnews.pro/news/dognavy-cybergym-technical-report.jsonld"}}