Accessibility Toolkit 1.0.0

• Outcome: Review
• Latest audit: May 28, 2026 at 9:37 AM UTC
• Version:
  1.0.0

Overview

The skill is accessibility-focused and not deceptive, but it gives under-scoped guidance for smart-home actions that can affect physical security.

Install only if you will adapt the templates manually. Keep confirmations or strict allowlists for locks, doors, alarms, medical routines, and other high-impact actions, replace example access codes with placeholders or secret storage, and avoid letting an agent mine conversation history without explicit consent and privacy limits.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (3)

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The example voice triggers are short, natural phrases like "Help" and "Movie time," which are prone to accidental activation from ordinary conversation, media playback, or other household speech. In this skill's context, the risk is elevated because the guidance explicitly favors low-friction execution and says not to require confirmation for reversible actions, increasing the chance that unintended commands will be acted on immediately.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: The arrival automation unlocks the front door automatically when presence detection says the person has entered the home zone, but the skill does not warn about geofencing inaccuracies, spoofed presence signals, shared-device risks, or the safety implications of autonomous unlocking. In an accessibility-focused skill, convenience pressure makes this more dangerous because users may deploy it as-is for a high-impact physical security control.

Ssd 3

Medium

• Confidence: 99% confidence
• Finding: The documentation includes a literal backup door code in plain text, which normalizes storing and exposing sensitive credentials directly in user-facing messages or repository content. If copied into a real deployment, this can leak an access secret through logs, screenshots, version control, or agent responses, enabling unauthorized entry.

VirusTotal

66/66 vendors flagged this skill as clean.