System Monitor

Security checks across malware telemetry and agentic risk

Overview

This is a static system-monitor dashboard with no hidden execution or data theft, but its displayed metrics are mock values unless the user wires it to real telemetry.

Treat this as a visual demo until it is connected to authenticated, real telemetry. Replace the sample private HTTP URL, prefer HTTPS or same-origin routing, restrict the monitor and /healthz endpoint to trusted users, and do not rely on the displayed health values for operations without verifying the data source.

SkillSpector

By NVIDIA

Vulnerability Patterns

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
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Description-Behavior Mismatch

Medium

Confidence: 96% confidence
Finding: The skill is presented as a live system monitor for server health, gateway connectivity, response times, and activity logs, but the JavaScript only generates randomized values and static log content. In a monitoring context, fabricated operational data can mislead users into believing systems are healthy or active, causing delayed incident response and unsafe operational decisions.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The page includes a hardcoded HTTP endpoint pointing to a private LAN address without any user-facing disclosure that the skill may attempt network access. Even though the current code does not call the endpoint, embedding an undisclosed internal service URL can expose internal topology details and creates risk if later code begins making insecure cleartext requests to that host.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal