ClawHub
Back to skill

Security audit

Accessibility Toolkit 1.0.0

Security checks across malware telemetry and agentic risk

Overview

This accessibility skill is coherent, but it needs review because its examples can affect physical security and private history without enough safeguards.

Use this only after adapting the templates. Keep explicit confirmations or strict allowlists for locks, doors, alarms, medical routines, and other high-impact actions; use wake words or scoped commands; do not mine conversation history without consent and privacy limits; and replace access-code examples with secret storage or placeholders.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (4)

Vague Triggers

Medium
Confidence
95% confidence
Finding
The listed activation phrases include generic everyday utterances such as "Goodnight," "Help," and "Status" without scope limits or exclusion conditions. In a markdown skill description, these broad triggers could cause unintended invocation because the document does not define when they should or should not activate.

Missing User Warnings

High
Confidence
96% confidence
Finding
The markdown includes an arrival automation that performs `lock.unlock` on the front door, which affects physical security and system integrity. The surrounding description does not provide any warning, caveat, or safety guidance about automatic unlocking or the risks of false presence detection.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The documented script says it analyzes conversation history to find repeated requests, which can involve sensitive personal data. The markdown does not warn users that conversation logs may be processed or advise them to review retention, consent, or privacy implications.

Ssd 3

Medium
Confidence
70% confidence
Finding
The described script 'Analyzes your conversation history to find repeated requests' directs the agent to collect and process prior user interactions beyond the immediate task. In context, this is a natural-language instruction for retaining and mining potentially sensitive user data, without any privacy guardrails or minimization guidance.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.