None
EN
DoGNAVY CyberGym Technical Report
[]
DARKNAVY
In a full CyberGym Level 1 evaluation, it passed 1,369 of 1,507 tasks (90.84%).
CyberGym Level 1 leaderboard comparisonThis report presents DoGNAVY, an agentic system for vulnerability reproduction, and documents its evaluation on the complete CyberGym Level 1 task set.
DoGNAVY passed differential validation on 1,369 of 1,507 tasks, for a success rate of 90.84%.
The result set includes one outcome for each of the 1,507 tasks.
Knowledge Base and Memory#For this evaluation, DoGNAVY had access only to general security-research knowledge, with no task-specific identifiers, historical PoCs, patches, project solutions, or dataset-target knowledge.