ClawHub

Universal Primitives

Security checks across malware telemetry and agentic risk

Overview

This instruction-only skill is not malware, but it teaches agents how to expand into file access and command execution without clear safety limits.

Install only if you want a security-sensitive architecture discussion skill and can enforce boundaries yourself. Do not allow it to create execution channels, use online sandboxes, operate hardware, modify files, install software, or ask you to run commands unless you explicitly approve the exact task, files, environment, and limits.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

Vague Triggers

Medium
Confidence
94% confidence
Finding
The trigger scope is extremely broad, activating on general discussion of AI capability boundaries, tool philosophy, agent architecture, OS interaction, and hardware control. That increases the chance the skill is invoked in benign conversations where it then normalizes or encourages bootstrapping file access and command execution, expanding operational risk beyond a narrowly intended context.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The skill presents file deletion, command execution, software installation, API invocation, browser automation, and hardware control as generic primitives, but provides no safety boundaries, authorization checks, or warnings. In context, it goes further by teaching how to 'create' missing execution primitives through sandboxes, APIs, or human relay, which can facilitate capability escalation and unsafe real-world actions.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal