Security audit

Ai Literacy Expert V4.3

Security checks across malware telemetry and agentic risk

Overview

This skill coherently generates AI-literacy teaching materials, courseware, games, and lesson-prep documents, with some privacy and deployment details users should review before using generated outputs in production.

Before installing or using this skill in a real school or training environment, review generated HTML and server code for CDN use, API submission, audit logging, localStorage drafts, and document downloads. Avoid entering student personal data, add privacy notices and retention limits for any logging, and keep API keys server-side as the skill recommends.

SkillSpector

By NVIDIA

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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 (5)

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The game capability uses broad trigger phrases such as “游戏”“闯关”“冒险”“玩中学”, which can match ordinary educational or conversational requests and route users into a more complex behavior than intended. In a skill that can generate executable single-file HTML artifacts, ambiguous routing increases the chance of surprising outputs, scope creep, and misuse of the more powerful game-generation path without clear user consent.

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The lesson-preparation triggers include broad requests like “帮我备课”“出一套教案”“生成题目”, which are common everyday phrases and may activate a workflow that generates multiple downloadable documents and packaged outputs. Because this skill supports one-click ZIP packaging and rich document generation, loose activation criteria can cause unintended data processing, overcollection during the Q&A flow, and unexpected artifact generation.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill explicitly lists observability and audit logging for user behavior, but provides no user-facing notice, consent flow, retention policy, or data-minimization constraints. In an education-oriented skill handling teachers and students, behavioral logging can expose sensitive usage patterns or educational content, making the privacy risk more significant in context.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The logging standard explicitly captures user input, delivery outputs, and adoption/usage metrics, but it does not require user notice, consent, retention limits, or minimization controls. In an education context involving teachers and potentially student-related lesson content, this can lead to privacy violations, over-collection, and secondary exposure of sensitive instructional or personal data through audit pipelines.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The guide explicitly proposes an AI API mode that sends audience, module, and granularity inputs to a server endpoint, but it does not require any user-facing notice, consent, or data-handling disclosure. In an education context, these inputs can reveal teaching plans, institutional context, or other potentially sensitive workflow data, so silent transmission creates a real privacy and transparency risk.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.