ClawHub
Back to skill

Security audit

智能内容归档

Security checks for vulnerabilities and agentic risk

Overview

This is mostly a content-archive skill, but it uses broad file/command authority and includes unrelated activation wording that could make it run in contexts beyond archiving.

Review this skill before installing. Use it only for explicit archive or archive-search tasks, avoid feeding private or authenticated URLs unless you are comfortable with local storage and possible LLM/API processing, and be cautious about granting exec access unless the platform can constrain commands to the archive workflow.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Description-Behavior Mismatch

Medium
Confidence
93% confidence
Finding
The skill is presented as an archive/search tool, but the description broadens invocation to project management, task planning, progress tracking, and team collaboration. This can cause the agent to invoke a tool with file access and command execution in contexts unrelated to archiving, increasing the chance of unintended data collection, storage, or processing of sensitive workspace content.

Vague Triggers

Medium
Confidence
95% confidence
Finding
The skill includes an unrelated, overly broad trigger ('use when project management, task planning, progress tracking, team collaboration') that can cause unintended invocation. In this skill's context, that is risky because the tool has read/exec/glob/grep capabilities and is designed to capture external content into persistent local storage, so accidental activation could process or retain data far beyond user expectations.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The description does not clearly warn users that archiving may fetch external URLs, extract full content, and send content through semantic/LLM processing. This undermines informed consent and can expose sensitive URLs, documents, or derived content to network services or local persistent storage without the user's clear understanding.

Static analysis

No suspicious patterns detected.