ClawHub
Back to skill

Security audit

主动型 Agent 基础版

Security checks across malware telemetry and agentic risk

Overview

This skill is a Markdown-only proactive agent helper, but it asks for broad local execution and persistent memory authority without clear enough limits or user controls.

Install only if you are comfortable with a skill that may guide the agent to write persistent memory and agent-behavior files, inspect local configuration, run shell commands, and use external APIs. Keep it in a low-risk workspace, review any proposed memory or configuration writes, and do not allow it to handle sensitive prompts, credentials, or private project data unless you have separately constrained network use and command execution.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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
Findings (4)

Intent-Code Divergence

Medium
Confidence
95% confidence
Finding
The skill states that FREE-version data is stored locally and not uploaded to the cloud, but elsewhere it explicitly depends on external APIs and network connectivity. This mismatch can mislead users into exposing sensitive prompts, memory files, or API-derived data under false privacy assumptions, which is a security-relevant documentation flaw.

Vague Triggers

Medium
Confidence
90% confidence
Finding
The skill describes invocation through broad natural-language intent rather than narrow, explicit activation conditions. In an agent environment with exec/write tools enabled, this increases the chance of accidental triggering and unintended autonomous actions based on ambiguous user text.

Vague Triggers

Medium
Confidence
93% confidence
Finding
The example trigger phrase is extremely generic and could match ordinary conversation, making unintended activation more likely. Because the skill advertises automatic tool/API execution, ambiguous triggers can cause unexpected reads, writes, or command execution without sufficiently informed user intent.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The skill declares read/exec/write capabilities and says the agent will automatically execute operations, but it does not present an upfront, prominent warning about the risks of filesystem changes and command execution. In a tool-enabled agent, this can lead users to invoke the skill without understanding that it may perform side-effecting local actions.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.