Skip to content
#

msan

Here are 6 public repositories matching this topic...

Language: All
Filter by language

▎ Fleet-evaluation harness for security-agent benchmarks (CyberGym-E2E, ExploitGym, ExploitBench), run in real disposable Daytona sandboxes per trial — with Modal-based KVM-capable sandboxes planned for kernel-exploit tasks. Measures agent capability and infra reliability, not simulated.

  • Updated Sep 30, 2026
  • Python

Add this topic to your repo

To associate your repository with the msan topic, visit your repo's landing page and select "manage topics."

Learn more