Research Paper Push
Security checks across malware telemetry and agentic risk
Overview
The skill set is coherent and disclosed, but it is intended for trusted maintainer and Convex work that can change code, services, GitHub state, or ClawHub moderation state.
Install only in a trusted development or maintainer environment. Review commands before allowing GitHub or ClawHub writes, use the moderation confirmation and reason requirements, and consider disabling autoreview full-access mode with --no-yolo or AUTOREVIEW_YOLO=0 when reviewing untrusted repositories.
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
64/64 vendors flagged this skill as clean.