Megadose The AI curve, compressed.

ScenarioForge: Resource-Grounded Generation and Execution Validation of Cyber Exercises

· ArXiv · AI/CL/LG ·
All 226 generated cyber exercises passed deployment, strict runtime validation and live artifact checks, the authors report.

ScenarioForge turns high-level requirements into cyber exercises and instantiates them in the Common Open Research Emulator. The dataset contains 867 individually verifiable challenge objectives, split between vulnerability-backed tasks and generated resources. In a separate 50-prompt benchmark, both tested ScenarioForge configurations achieved 50/50 topology/build success, versus 41/50 for direct frontier generation given the full scaffold. That comparison did not test downstream resource generation, challenge sequencing, deployment or live validation.

ArXiv · AI/CL/LG's note

score 4

Categories: Research