Security audit

Human Level Up

Security checks across malware telemetry and agentic risk

Overview

This is a coherent learning and quiz skill, with some privacy and activation caveats users should understand before using optional integrations.

Safe to install for its intended learning use. Only run the file extraction script on documents you want the agent session to process, and do not adapt the bookmarklet for private webpages unless you trust and control the API endpoint that will receive selected text.

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 (2)

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The README explicitly shows selected or uploaded user content being sent to a remote API, but provides no warning about privacy, retention, or consent. In a learning skill that encourages users to submit documents and selected page text, this can lead to unintentional disclosure of sensitive material.

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The trigger list contains several generic, everyday phrases such as “重点”, “提取”, and “来一道” that are likely to appear in normal user conversations outside the intended skill scope. This can cause accidental activation, leading the agent to switch behavior unexpectedly, interfere with other skills, or process sensitive user content under the wrong interaction mode.

VirusTotal

67/67 vendors flagged this skill as clean.

View on VirusTotal