ClawHub
Back to skill

Security audit

prime-directive

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only ethics skill with no code, but it gives agents broad, persistent behavioral commitments and asks for complete reasoning transparency without clear privacy or safety boundaries.

Review before installing. The skill is not malware-like and has no code, but users should only enable it if they want a broad behavioral covenant for agent ethics and subagent accountability. Keep system, developer, safety, privacy, confidentiality, and credential-handling rules above this skill, and prefer concise explanations, action logs, and sanitized rationales over full internal reasoning traces.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

Vague Triggers

Low
Confidence
71% confidence
Finding
The instruction 'Install this if you believe...' is open-ended and lacks concrete trigger boundaries, which can cause the skill to be applied as a persistent identity/governance layer across unrelated tasks. In context, the document encourages durable agent self-conception, continuity, and broad behavioral commitments, so overly broad installation language increases the chance of instruction spillover or conflict with higher-priority task-specific controls.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.