ClawHub
Back to skill

Security audit

用模式转换

Security checks for vulnerabilities and agentic risk

Overview

This text-processing skill is not clearly malicious, but it asks for broad write and command-execution authority that is not well scoped for its stated purpose.

Install only if you are comfortable granting this skill file-write and command-execution capability. Use it in a constrained workspace, review any proposed command or file change before allowing it, and avoid providing sensitive documents or API keys unless the publisher clarifies exactly what is sent to external services.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (4)

Context-Inappropriate Capability

Medium
Confidence
91% confidence
Finding
The skill is presented as a text-processing utility, but the documentation also advertises file handling, external API integration, and system command execution capabilities that are not narrowly justified by that purpose. This mismatch expands the effective privilege and attack surface of the skill, making it easier for a user or downstream agent to invoke risky behaviors under an innocuous label.

Intent-Code Divergence

Medium
Confidence
88% confidence
Finding
The documentation states the skill is local/offline, yet nearby sections describe LLM API dependence, API key configuration, and an MD+execute() classification. These conflicting claims can mislead operators into granting trust or permissions under the false assumption that no network or execution side effects occur.

Vague Triggers

Medium
Confidence
79% confidence
Finding
The description uses very broad 'use when' language covering content generation, optimization, automation, and multiple workflows without clear trigger boundaries. In an agent environment, vague invocation criteria increase the chance the skill is auto-selected in contexts where its read/write/exec tooling could be unnecessarily exposed or misused.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The markdown advertises file writing, external service calls, and command execution, but the document does not place prominent user-facing warnings or consent requirements near those capabilities. In a skill that appears primarily text-oriented, this can cause users or orchestrators to underestimate system-impacting behavior and permit destructive or data-exfiltrating actions.

Static analysis

No suspicious patterns detected.