What was done
SentinelLABS used the fast16 sample to see how far frontier models get on their own in a defensive analysis.
Where the human stepped in
Not reported yet. If this is your project, tell us what the AI got wrong and how you fixed it.
Malware (defensive)Research
Target: fast16 malware
Eight-stage malware analysis to test long-horizon agents
by SentinelLABSRead the writeup
Related projects
Malware samples (Cerber, Hydra, NetWire, PolyRansom)Wrote reusable Codex CLI skills for malware unpacking and IOC extraction in FLARE-VMmfc42ul.dll (publicly documented trojan sample)Built a structured agentic malware-analysis pipeline and tested it on a known government trojanMalware samples (lab analysis)Tested GPT and Claude models as malware analysts with Ghidra, x64dbg and REMnux MCP servers