ClawHub
Back to skill

Security audit

学术论文阅读技能 (Paper Reader)

Security checks across malware telemetry and agentic risk

Overview

This is a markdown-only academic paper-reading skill with a broad automatic workflow, but its behavior is disclosed and aligned with analyzing user-provided papers.

Install this if you want a Chinese-oriented assistant for deep academic paper reading. Be explicit when you only want a short answer, one section analyzed, or output in another language, because the skill is written to process a provided paper broadly and automatically.

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 (3)

Vague Triggers

Medium
Confidence
92% confidence
Finding
The skill description covers very broad triggers such as providing a PDF path, asking about paper content, or explaining a term or paragraph, which can cause the skill to activate for ordinary academic questions without clear user intent to invoke the full workflow. Over-broad activation increases the chance of unintended file access, excessive processing, and surprising behavior that overrides a narrower user request.

Vague Triggers

Medium
Confidence
95% confidence
Finding
The instruction to automatically enter the full workflow whenever a paper file is provided lacks guardrails on scope, consent, and task matching. This can lead to unnecessary full-document translation or analysis when the user only wanted a summary, a specific answer, or a limited excerpt, creating privacy, cost, and least-privilege concerns.

Natural-Language Policy Violations

Medium
Confidence
84% confidence
Finding
Mandating a Chinese output artifact without offering language choice can conflict with user intent and reduce transparency or usability, especially in multilingual environments. While not a direct code-execution risk, it is a policy and safety issue because it may transform or disclose content in an unexpected form and make review harder for users expecting the original language.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.