ClawHub
Back to skill

Security audit

笔记

Security checks for vulnerabilities and agentic risk

Overview

This note-taking skill is not clearly scoped and includes task-command, export, and persistent logging instructions that do not fit its stated note-formatting purpose.

Review this skill before installing. It appears template-like and not malicious, but its instructions are broad and inconsistent; avoid using it with sensitive notes unless command use, file locations, exports, and logging are clarified or removed.

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 (4)

Intent-Code Divergence

Medium
Confidence
94% confidence
Finding
The documented interface describes CLI behaviors such as printing full data files and logging all actions, which conflicts with the skill's stated note-formatting role and declared request schema. This inconsistency can cause an agent to perform unintended read/write/export actions, increasing the chance of sensitive note content being exposed or mishandled.

Intent-Code Divergence

Medium
Confidence
96% confidence
Finding
The examples instruct the agent to use task-management commands like add, done, priority, remind, and export, which do not match the note-taking skill's declared function. In an agentic environment with exec/write tools enabled, contradictory examples can steer the agent into executing unintended commands or creating/modifying files outside the user's expectation.

Vague Triggers

Medium
Confidence
85% confidence
Finding
The invocation description is overly broad, stating the skill should be used for AI model calls, intelligent dialogue, agent orchestration, and LLM applications in general. Such vague scope encourages over-invocation and tool use outside the narrow note-taking context, which can lead an agent to access files, execute commands, or process sensitive data unnecessarily.

Ssd 3

Medium
Confidence
97% confidence
Finding
The skill states that full data contents may be printed to stdout and that all actions are logged to history.log for auditing. In a note-taking context, notes can contain sensitive personal, business, or credential-adjacent information, so unconditional full-output display and persistent logging materially increase the risk of data leakage through terminals, logs, transcripts, or downstream integrations.

Static analysis

No suspicious patterns detected.