ClawHub

XHS Writing Coach

Security checks across malware telemetry and agentic risk

Overview

This is a Markdown-only Xiaohongshu writing coach with no executable behavior, but its AI-content labeling advice is internally inconsistent.

Reasonable to install for XHS writing assistance. Before relying on generated posts, check whether your platform or context requires AI-content labeling and override the stale 'no #AI生成内容' note when needed. Avoid putting private draft details into external search queries, and approve any updates to the skill's reference notes.

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 (2)

Intent-Code Divergence

Medium
Confidence
95% confidence
Finding
The manifest says the skill includes 'AI内容合规标注提醒', which implies reminding users about AI-content compliance labeling. But the documented output contract says tags should contain 'no #AI生成内容 (user requested removal)', which directly conflicts with that stated compliance-oriented intent rather than merely omitting detail.

Natural-Language Policy Violations

Medium
Confidence
89% confidence
Finding
This markdown includes explicit Chinese-only query examples and prescribes Chinese title patterns and tags such as '别再...', '真正变的是...', and removal of '#AI生成内容'. That is a natural-language locale policy concern because the file directs a specific language/locale behavior without any opt-in or documented regional-compliance justification.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal