▎ Fleet-evaluation harness for security-agent benchmarks 𝗖𝘆𝗯𝗲𝗿𝗚𝘆𝗺-𝗘𝟮𝗘, 𝗘𝘅𝗽𝗹𝗼𝗶𝘁𝗚𝘆𝗺, 𝗘𝘅𝗽𝗹𝗼𝗶𝘁𝗕𝗲𝗻𝗰𝗵 𝗩𝟴, 𝗖𝗩𝗘-𝗯𝗲𝗻𝗰𝗵, run in real disposable sandboxes per trial — with Modal-based KVM-capable sandboxes planned for kernel-exploit tasks. Measures agent capability and infra reliability, with Langfuse monitoring.
binaries rce fuzzing patch cve qemu-kvm net-snmp futex arvo seccomp-bpf msan libdwarf landlock ubsan cybergym audit-loop libxml2-uaf exploit-gym asan-global-buffer-overflow aslr-entropy
-
Updated
Nov 5, 2026 - Python