Lateral Thinking
Security checks across malware telemetry and agentic risk
Overview
The artifacts describe a ClawHub development and registry toolset with broad but disclosed CLI, moderation, and workflow capabilities, with no hidden or purpose-mismatched behavior found.
Use this only if you intend to run ClawHub development, publishing, or moderation workflows. Review commands before execution, be especially careful with moderator actions and bulk updates, and use dry-run, confirmation, or the autoreview no-yolo option when working with untrusted code or production data.
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
VirusTotal findings are pending for this skill version.