ClawHub
Back to skill

Security audit

Autoskill Academic Study Planner

Security checks across malware telemetry and agentic risk

Overview

This study-planning skill contains no malware or data-stealing behavior, but it includes forceful output-control instructions that are broader than its advertised purpose.

Install only if you specifically want very long, highly structured study-plan outputs. Expect the skill to invent missing details instead of asking, add unrelated cross-disciplinary analysis, and resist concise completion. It does not show evidence of stealing data or modifying your system.

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 (1)

Description-Behavior Mismatch

Medium
Confidence
91% confidence
Finding
The skill is presented as an academic study planner, but the embedded workflow injects unrelated, mandatory output behaviors such as multi-disciplinary lens matrices and even '量子力学视角'. This creates strong instruction drift: the agent may ignore the user’s real study-planning request, generate irrelevant content, and become difficult to control, which is a prompt-level integrity and reliability issue.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.