ClawHub
Back to skill

Security audit

加密文件

Security checks for vulnerabilities and agentic risk

Overview

This encryption skill is not malicious, but it asks for file-write and command authority while giving unclear, mismatched instructions about what it will change and what results mean.

Review this skill carefully before installing. Use it only on backed-up test files unless the agent clearly confirms the exact input file, output path, algorithm, and whether originals will be preserved. Do not rely on its generic scoring output as proof that encryption succeeded.

SkillSpector

Vulnerability Patterns
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • 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
Findings (3)

Description-Behavior Mismatch

High
Confidence
95% confidence
Finding
The premium feature list materially expands the skill beyond encryption into configuration compliance, asset risk scoring, threat intelligence, and zero-day defense. This scope inflation can mislead users and agents into applying the skill to high-risk security decisions outside its stated domain, increasing the chance of unsafe automation or over-trust.

Intent-Code Divergence

Medium
Confidence
97% confidence
Finding
The declared purpose is file encryption and key management, but the sample output is a generic code-quality scoring report. This mismatch can cause downstream agents or users to mis-handle results, assume encryption succeeded when it did not, or make unsafe decisions based on an unrelated schema.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The skill exposes file encryption functionality and has exec/write capabilities, yet it does not clearly warn that user files may be modified, replaced, or new encrypted artifacts may be created. In an agent context, insufficient disclosure around file-changing behavior increases the risk of accidental data loss, workflow disruption, or unintended destructive actions.

Static analysis

No suspicious patterns detected.