ClawHub
Back to skill

Security audit

Markdown 工具箱

Security checks across malware telemetry and agentic risk

Overview

This Markdown helper is not malicious, but it asks for command execution and file-writing authority that is broader and less clearly bounded than its formatting purpose needs.

Review this before installing if your agent can run shell commands automatically. Use it for Markdown generation and validation only, and require confirmation before npm installs, markdownlint runs, ping/network diagnostics, or any file modifications outside the specific document you requested.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (3)

Context-Inappropriate Capability

Medium
Confidence
93% confidence
Finding
The skill declares `exec` capability for what is presented as a Markdown generation/validation tool, but the document does not tightly constrain what commands may be run or why shell access is strictly necessary. In an agent setting, unnecessary command execution expands the attack surface and could enable unsafe command invocation through prompt injection or overbroad task interpretation.

Vague Triggers

High
Confidence
90% confidence
Finding
The trigger condition says the skill should be used for broad categories like code generation, programming assistance, debugging, testing, and deployment, which far exceed a Markdown toolkit's stated purpose. Overbroad activation criteria can cause the agent to invoke a tool with `exec`/`write` privileges in unrelated software-development contexts, increasing the chance of unsafe actions or privilege misuse.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill advertises `exec` and `write` capabilities in metadata, but the user-facing description does not prominently warn that commands may run or files may be modified. In agent environments, hidden or insufficiently disclosed side effects reduce informed consent and make it easier for an unrelated content task to become a command-execution or file-modification action.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.