ClawHub

SCI论文三轮滚动精修助手

Security checks across malware telemetry and agentic risk

Overview

This appears to be an academic writing/editing skill, but it explicitly promotes hiding AI involvement from detection in scholarly submissions.

Review before installing. Use only for legitimate clarity, structure, and language editing, not to hide AI assistance or bypass journal, school, or employer disclosure requirements.

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)

Ssd 2

Medium
Confidence
97% confidence
Finding
The skill explicitly frames one of its goals as removing 'AI writing traces,' which encourages evasion of AI-origin detection rather than ordinary academic editing. In an academic-paper context, this can facilitate misrepresentation of authorship or concealment of AI assistance, undermining institutional disclosure policies and integrity controls.

Ssd 2

Medium
Confidence
98% confidence
Finding
The user-targeting section markets 'anti-AI-trace' writing as a service benefit, which strengthens the operational intent to help users disguise machine-assisted text. Because the skill is aimed at SCI/core-journal submissions, the context increases risk by aligning the evasion with scholarly publishing workflows where disclosure and originality expectations are important.

Ssd 2

Medium
Confidence
99% confidence
Finding
The 'anti-AI detection standards' section provides concrete heuristics for identifying and rewriting features likely to trigger AI-detection suspicion, which amounts to procedural guidance for detector evasion. This is more dangerous in context because it transforms a normal editing assistant into a workflow for systematically laundering AI-generated text for academic submission.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal