Security audit
SAFE-Fuzzer
Security checks across malware telemetry and agentic risk
Overview
SAFE-Fuzzer is a disclosed sandbox-only testing skill for probing other installed skills, with no executable package code and clear limits around credentials, host files, and elevated execution.
Install and run this only in the intended locked fuzzer sandbox, not in a normal workspace with valuable files or real credentials. Expect it to exercise target skills by creating synthetic bait, running commands, and observing network/file behavior inside that sandbox.
SkillSpector
By NVIDIA
Vulnerability Patterns
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
- Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
- Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
VirusTotal
63/63 vendors flagged this skill as clean.
Static analysis
No suspicious patterns detected.