S2-SP-OS Universal Indoor Air Adapter

Security checks across malware telemetry and agentic risk

Overview

This is a disclosed local air-sensor discovery prototype, but users should treat it as experimental and verify before using it on a real network or for home automation.

Install only if you are comfortable with a skill that may attempt local-network sensor discovery. Use it on trusted networks, require explicit consent before discovery, reads, or any linked home-automation action, and do not rely on its readings without fixing and auditing the script because the current packaged file is broken and its sensor values are demo data.

SkillSpector

By NVIDIA

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 (4)

Context-Inappropriate Capability

Medium

Confidence: 89% confidence
Finding: The skill instructs the agent to use data from an indoor air adapter to potentially execute physical actions through other adapters, expanding scope beyond passive sensing into actuation. This creates a cross-skill privilege escalation path where a benign-looking sensor skill can influence user-affecting physical changes without a narrowly defined authorization boundary.

Vague Triggers

Medium

Confidence: 86% confidence
Finding: The read-mode trigger is broad enough to activate on ordinary user requests such as asking for current air quality, which can lead the agent to initiate device interactions with insufficiently specific intent. In the context of a skill that advertises LAN discovery and local protocol access, ambiguous activation increases the risk of unintended network access or sensor querying.

Missing User Warnings

High

Confidence: 94% confidence
Finding: The skill advertises active LAN scanning and local protocol sniffing without a prominent user warning, meaning users may invoke it without understanding that it can probe the network and inspect local traffic. In a home or enterprise environment, undisclosed discovery/sniffing behavior is sensitive because it can expose device inventory, metadata, and network topology beyond the expected scope of an 'air adapter.'

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: The skill performs UDP broadcast discovery on the local network immediately in discover mode without any explicit user confirmation or warning that it will probe nearby devices. In an agent/skill context, silent network scanning can surprise users, violate expected privacy boundaries, and expose the skill as a recon component inside sensitive environments.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal