Security audit

Accessibility Toolkit

Security checks for vulnerabilities and agentic risk

Overview

The skill is accessibility-focused, but it recommends unsafeguarded smart-home and voice automations that could affect physical security without clear user control.

Review this skill carefully before installing. It is not deceptive and does not ship executable code, but its examples should be treated as unsafe defaults unless you add explicit opt-in controls, stronger presence verification, confirmations for locks and other sensitive actions, and clear privacy limits for any conversation-history analysis.

A.I.G

Vulnerability Patterns

Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally

Findings (2)

T09 · Insecure Skill Coding Practices

Warning

Location: SKILL.md:35
Finding: Unconditional Confirmation Bypass for Reversible Actions
Content: View full analysis
Remediation: View remediation

T09 · Insecure Skill Coding Practices

Error

Location: SKILL.md:82
Finding: Exterior Door Automatically Unlocks from a Single Zone-Presence Event
Content: View full analysis
Remediation: View remediation

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
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
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 98% confidence
Finding: The arrival automation unlocks the front door automatically on zone entry, but the skill text does not warn about the security implications or recommend safeguards. Presence detection can be spoofed or misfire, and in a smart-home setting this could expose the home to unauthorized physical access.
Content: No source excerpt is available for this finding.

Autonomous Decision Making

Medium

Category: Excessive Agency
Confidence: 91% confidence
Finding: The skill advocates immediate execution with no confirmation for voice-triggered actions, including a broad emergency-style command. Even if some actions are reversible, voice systems are error-prone, and autonomous execution can lead to unintended environmental changes or false alerts in a high-dependency accessibility setting.
Content: Scanner excerpt · SKILL.md (reported line 32)May include surrounding context.
"Goodnight" → Bedtime scene, lock doors, set thermostat, silence notifications "I'm working" → Focus mode, desk lights, DND, close distracting tabs "Movie time" → Dim lights, TV on, adjust audio "Help" → Immediate attention, no confirmation dialogs

text
**Never require confirmation for reversible actions.** Just do it. They can say "undo" if wrong.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 96% confidence
Finding: The trigger phrase "Help" is dangerously generic in a voice-first skill and can be invoked by normal conversation, media playback, or bystanders. In this accessibility context, the command is explicitly mapped to immediate action with no confirmation, which raises the chance of accidental emergency behaviors or unintended automations.
Content: No source excerpt is available for this finding.

Autonomous Decision Making

Medium

Category: Excessive Agency
Confidence: 75% confidence
Finding: Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
Content: Scanner excerpt · SKILL.md (reported line 151)May include surrounding context.
✓ Doors locked ✓ Thermostat 68°F

No confirmation needed — all reversible with one word.

text
## Scripts

Ssd 3

Medium

Category: Not specified by scanner
Confidence: 85% confidence
Finding: The friction audit script states it analyzes conversation history to identify repeated requests, implying retention and post-processing of potentially sensitive user interactions. Because this skill targets users with disabilities, those conversations may contain health, routine, and household data, making privacy exposure more consequential.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.