LG ThinQ

Security checks across malware telemetry and agentic risk

Overview

This LG ThinQ skill is mostly coherent, but it deserves Review because it can send unrestricted raw control commands to real appliances using the user's account token.

Install only if you trust the publisher and are comfortable giving the agent ThinQ appliance-control access. Protect the token file, verify the thinqconnect package before installing it, and avoid using the raw command unless you personally review the exact payload and target device.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration

Findings (5)

Description-Behavior Mismatch

Medium

Confidence: 95% confidence
Finding: The script advertises appliance-specific operations, but also exposes a generic raw command interface that accepts arbitrary JSON and forwards it directly to the ThinQ control API. This bypasses the narrow safety boundaries implied by the skill description and allows any caller with access to the script and token to issue undocumented or unintended commands to supported devices.

Description-Behavior Mismatch

Medium

Confidence: 96% confidence
Finding: The raw command path resolves an arbitrary device identifier and sends attacker-controlled JSON to that device without restricting device class to the manifest's fridge/washer/dryer/AC scope. In the context of a user-facing agent skill, this expands authority beyond what users and reviewers would reasonably expect, increasing the chance of unsafe control over other LG devices on the same account.

Vague Triggers

Medium

Confidence: 83% confidence
Finding: The activation wording is broad enough that ordinary conversation about LG appliances could invoke the skill unexpectedly. In a skill that can control real devices, over-broad triggering raises the risk of unintended execution of status checks or state-changing actions on household appliances.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The documentation describes commands that can change real appliance settings without a clear warning that they have immediate physical-world effects. That omission is risky in context because refrigerator, freezer, and AC settings can affect food safety, energy use, and household operation if triggered accidentally or misunderstood.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The raw control feature performs immediate arbitrary device actions with no warning, policy check, or user confirmation step. For physical-device control, lack of confirmation materially increases the risk of unintended or unsafe actions, especially when the command can affect any reachable ThinQ device and the payload is unrestricted.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal