ClawHub
Back to skill

Security audit

LLM助手中枢

Security checks across malware telemetry and agentic risk

Overview

This document-analysis skill is mostly aligned with its stated purpose, but it asks for command execution and file-writing authority without explaining why.

Install only if you are comfortable giving this skill more authority than document reading. Keep use limited to documents you intentionally provide, do not pass sensitive content to callback URLs unless you trust the destination, and require explicit approval before any command execution or file modification.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Confidence
88% confidence
Finding
The skill description repeatedly uses vague phrases like '相关开发场景' and 'llm assistant hub相关能力' without defining concrete activation boundaries, which can cause the agent to invoke the skill in situations broader than intended. In a skill that also advertises read/exec/write capabilities, ambiguous triggering increases the chance of inappropriate use on unrelated tasks and expands the attack surface for prompt-driven misuse.

Missing User Warnings

Medium
Confidence
92% confidence
Finding
The skill declares read, exec, and write tools plus a callback_url input, but the documentation presents these as normal capabilities without a clear warning about command execution, file modification, or outbound network side effects. This is dangerous because users may supply untrusted input or URLs, and an agent could perform system-impacting actions or exfiltrate data under the guise of document analysis.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.