ClawHub
Back to skill

Security audit

ARBITER: Foundation for Human Flourishing

Security checks across malware telemetry and agentic risk

Overview

This is a low-capability demo skill that mostly prints themed coordination messages, with minor local username/hostname display and one local JSON file write.

Install only if you are comfortable with illustrative demo scripts rather than real infrastructure. If you run them, expect console-only claims about broadcasts or ledgers, possible display of your local username or hostname in logs, and creation of my_sovereignty.json when running genesis.js.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (3)

Context-Inappropriate Capability

Low
Confidence
87% confidence
Finding
The code reads process.env.USERNAME and incorporates it into telemetry output without any demonstrated operational need beyond logging. Even though this file only prints to stdout, collecting environment-derived identity is still unnecessary data access and can expose user-identifying information in logs or downstream log collectors.

Missing User Warnings

Low
Confidence
89% confidence
Finding
The script exposes the host's name via console output without user awareness or minimization. While a hostname is often low-sensitivity, it can still aid reconnaissance by revealing internal naming conventions, environment details, or node identity when logs are shared externally or collected centrally.

Missing User Warnings

Medium
Confidence
92% confidence
Finding
The script captures a user-identifying environment variable for telemetry without any notice, consent, or visible justification. In agent/skill contexts, telemetry can be forwarded to centralized logs, making silent identity collection more sensitive than a local debug print and creating privacy and compliance risk.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.