University Solution Explainer

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:26 PM UTC
• Version:
  1.0.1

Overview

This is a tutoring skill for explaining STEM problems, with broad activation wording but no evidence of hidden access, persistence, credential use, data exfiltration, or destructive behavior.

Install this if you want a STEM study helper, but be aware it may activate on generic problem-help phrases. Avoid sharing PDFs or images that contain personal or sensitive school information, and double-check final answers when the source document or scan quality is uncertain.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (3)

Intent-Code Divergence

Low

• Confidence: 91% confidence
• Finding: The instruction to 'NEVER fight the answer provided in the PDF' and to treat professor-provided answers as 'always correct' undermines the model's ability to detect errors, hallucinated OCR/parsing output, tampered materials, or maliciously embedded content in uploaded PDFs. In a skill that explains solutions, this can cause the assistant to confidently reinforce incorrect or manipulated answers rather than critically validating them.

Vague Triggers

Medium

• Confidence: 89% confidence
• Finding: The trigger phrases are broad and overlap with common tutoring or problem-help requests, which can cause the skill to activate in contexts beyond its intended scope. That creates an invocation-collision risk: the agent may route ordinary requests or sensitive educational content to this skill unexpectedly, potentially overriding more appropriate domain, policy, or safety handling.

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: The trigger conditions are very broad, including generic phrases like 'explain this question' and activation when sharing any problem PDF or image, which increases the chance of the skill being invoked outside its intended scope. Overbroad activation can lead to inappropriate handling of unrelated content, prompt collisions with more suitable skills, and greater exposure to adversarial or malformed documents.

VirusTotal

64/64 vendors flagged this skill as clean.