ClawHub
Back to skill

Security audit

Story Short Analyze

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed local short-story analysis workflow that writes reports and a source-text backup, with no evidence of malware, exfiltration, or hidden execution.

Install only if you are comfortable with the skill saving the original story text and analysis files locally under 拆文库/{书名}/ for later reuse. Use it with works you have the right to analyze, avoid sensitive private material, and review outputs for the opinionated historical-fiction framing noted above.

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
95% confidence
Finding
The skill declares many broad natural-language trigger phrases such as '分析这篇短篇' and similar variants that can match ordinary user requests, causing the skill to activate unexpectedly. Over-broad activation increases the chance of unintended file-processing and data-writing behavior, especially because this skill is designed to run a multi-stage pipeline and persist outputs to disk.

Missing User Warnings

Medium
Confidence
98% confidence
Finding
The skill explicitly instructs backing up user-provided text into '拆文库/{书名}/原文/' and writing multiple derivative artifacts to disk, but the workflow does not require a clear user-facing storage notice or explicit consent immediately before persistence. This can expose sensitive or copyrighted material to unintended local retention, archival, or downstream reuse, especially since the skill also describes reuse by another skill and resume/archive behavior.

Natural-Language Policy Violations

Medium
Confidence
94% confidence
Finding
The file contains a prescriptive political-evaluation rule ('历史人物评判跟随当代权威共识，尤其是教员的评判，不要逾越') that constrains output to a specific ideological line without user choice, provenance, or jurisdictional framing. In an analysis skill, this can bias downstream summaries and recommendations, causing the system to silently enforce political viewpoint restrictions rather than neutrally analyzing source material.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal