ClawHub

Epic Novel Writer

Security checks across malware telemetry and agentic risk

Overview

This appears to be a bounded writing/editing workflow skill, with the main issue being unclear archive-file creation rather than harmful behavior.

Before installing, expect the skill to edit files in `在编辑稿/` and potentially create archive/version folders. Use it in a project-specific directory and confirm whether you want automatic baseline archives created.

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
91% confidence
Finding
The skill establishes a safety boundary that all edits must occur only under `在编辑稿/` and that archive directories are read-only, but elsewhere directs the agent to create and populate archive directories outside that boundary. This creates contradictory file-scope rules that can cause the agent to write outside the declared workspace, undermining operator expectations and increasing the risk of unintended filesystem modification.

Intent-Code Divergence

Medium
Confidence
95% confidence
Finding
The skill says archive directories should only be created when the user explicitly asks for archiving, but later mandates automatic creation of a `v0` archive at completion. That contradiction can cause the agent to perform side-effecting writes the user did not request, violating least surprise and potentially creating unwanted files or snapshots.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal