ClawHub
Back to skill

Security audit

Markdown编辑工具

Security checks across malware telemetry and agentic risk

Overview

This Markdown skill has no embedded executable code, but it asks for command execution and describes broad API/file capabilities that are not clearly scoped to Markdown formatting.

Review this skill before installing. It appears to be a Markdown helper, but only use it where command execution is disabled or tightly controlled, and avoid providing API keys unless the publisher clarifies exactly why they are required and what commands or services may be used.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (4)

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The skill is presented as a Markdown formatting tool, but the documentation later claims generic file processing, API integration, and command execution capabilities. This creates scope ambiguity that can mislead users and agents into granting broader privileges than necessary, increasing the chance of unsafe invocation or abuse.

Context-Inappropriate Capability

Medium
Confidence
97% confidence
Finding
Declaring exec for a Markdown editing skill is dangerous because command execution is far more powerful than the stated use case requires. If an agent relies on this manifest, attacker-controlled content or ambiguous instructions could cause shell commands to run, leading to code execution, file tampering, or data exfiltration.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The description uses broad, vague applicability language without clear activation boundaries, making it easier for an agent or user to apply the skill outside its intended domain. That ambiguity becomes more dangerous because the skill also declares sensitive capabilities like write and exec, which could be invoked under weakly justified contexts.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The metadata exposes read, write, and exec capabilities without a prominent warning explaining that the skill can modify files and run commands. This undermines informed consent and increases the risk that users or orchestration systems will invoke the skill with more trust than its privileges warrant.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.