Homey

Security checks across malware telemetry and agentic risk

Overview

This skill does what it advertises: it controls Homey smart-home devices, including actions that can affect real locks, thermostats, lights, and automations.

Install only if you want Claw/Codex to have Homey control authority. Confirm exact device or flow names before running state-changing commands, especially unlocks, thermostat changes, appliances, and flow triggers. Protect Homey tokens, prefer prompt/stdin entry over shell history, and avoid using this skill on shared machines unless the config file permissions and token scope are acceptable.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
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
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (8)

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The examples explicitly show security-sensitive actions such as unlocking a door and triggering automations, but provide no warning, confirmation guidance, or safety caveats. In a smart-home control skill, these commands can cause immediate real-world effects affecting physical security, so normalizing them as routine examples increases the risk of unsafe or unintended use.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The AI integration section presents natural-language requests being translated directly into device-control commands without any warning that AI outputs can actuate physical devices. This is dangerous because users or downstream agents may treat the examples as safe defaults, enabling unintended lighting, climate, or other device changes from ambiguous, mistaken, or adversarial prompts.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: Advertising fuzzy matching for device control without strong cautions increases the chance that a typo, ambiguous phrase, or approximate match will control the wrong device. In a smart-home context, mis-targeting can have physical consequences, especially if similar names map to security, climate, or safety-related devices.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The documentation provides direct commands to turn devices on or off, change thermostat and lock state, and trigger automations, but does not warn that these actions cause immediate real-world changes in a user's home. In this context, the omission is significant because an agent may treat the examples as routine operations and perform unsafe actions affecting physical security, energy use, or occupant safety.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The testing guide instructs users to turn devices on/off, change capabilities, and trigger flows on a real Homey environment without an explicit warning that these actions will affect actual devices and automations. In a smart-home context, that can cause unintended physical actions or automation side effects during testing, especially if users copy commands verbatim against production homes.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The documentation explicitly includes commands that can turn devices on/off or set capabilities, but it provides no warning that these actions affect real smart-home hardware. In the context of a home automation skill, this increases the risk of accidental execution by users or downstream agents, potentially causing unintended physical-world effects such as unlocking, powering equipment, or changing environmental controls.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: Triggering a Homey flow can execute arbitrary automations, which may chain into multiple real-world actions beyond what the command name suggests. Because the docs omit any warning about these side effects, a user or agent may treat the command as low risk when it can activate alarms, locks, appliances, or other sensitive routines.

Missing User Warnings

Medium

Confidence: 86% confidence
Finding: The authentication examples show storing local API keys and cloud tokens, including via environment variables and shell pipelines, without warning about exposure in shell history, logs, process environments, or shared systems. In a smart-home context, compromised credentials can grant broad visibility and control over devices, zones, and automations.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal