Security audit

🗓️ 30天学习计划制定师

Security checks across malware telemetry and agentic risk

Overview

This study-planning skill should be reviewed because it can share sensitive student learning and emotional-profile details in parent-facing outputs without clear consent controls.

Install only if you are comfortable with the skill using historical learning-profile data and if students and guardians clearly agree on what can appear in family dashboards. Avoid enabling parent-facing emotional or diagnostic summaries unless there is explicit consent and a real need.

SkillSpector

By NVIDIA

Vulnerability Patterns

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
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Context-Inappropriate Capability

Medium

Confidence: 95% confidence
Finding: The skill explicitly instructs the agent to present student emotional and learning-difficulty diagnostics in a parent-facing dashboard, including anxiety, avoidance, and performance signals. In an education context involving minors, sharing this sensitive inferred data with parents without a strong consent and necessity model creates a real privacy and safety risk, even if phrased as supportive guidance.

Vague Triggers

Medium

Confidence: 89% confidence
Finding: The skill says it should be activated for broadly common study-planning requests and even says such scenarios should always invoke this skill. This increases unnecessary collection and use of historical learning data in situations where a simpler non-sensitive planning flow would suffice, expanding exposure of sensitive student information.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The description emphasizes use of 'real learning data' and family-facing planning outputs but does not prominently warn users that historical dialogue, behavioral patterns, and potentially sensitive student data may be accessed and summarized. That undermines informed consent and makes privacy-invasive operation more likely.

Ssd 3

Medium

Confidence: 92% confidence
Finding: The skill directs the model to retrieve and summarize sensitive student history such as error-frequency distributions, weak points, activity levels, and procrastination patterns in plain language. This is privacy-sensitive profiling data, and presenting it directly in routine responses increases risk of overexposure, especially if identity, audience, or consent are not carefully validated.

Ssd 3

High

Confidence: 98% confidence
Finding: The parent dashboard section explicitly tells the model to reveal sensitive diagnostic and emotional information to parents in softened language, including anxiety, avoidance, weak-point persistence, and engagement lapses. Softening the wording does not reduce the sensitivity of the disclosure; it may instead normalize sharing inferred mental-state data beyond the student, creating substantial privacy and trust harms.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.