Security audit
代码沙箱
Security checks across malware telemetry and agentic risk
Overview
This skill describes running untrusted code safely, but it does not provide or require an actual sandbox to enforce those protections.
Treat this as a sandbox design checklist, not a ready-to-use sandbox. Only install or use it for untrusted-code workflows if you already have a separate, verified isolated runtime with network blocking, filesystem limits, resource caps, and cleanup; otherwise use it for static review guidance only.
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
65/65 vendors flagged this skill as clean.