ClawHub

伐谋 - 实验结果渲染

Security checks across malware telemetry and agentic risk

Overview

This skill is a local visualization helper, with the main caution that generated pages load third-party scripts and fonts when opened.

Install only if you are comfortable with a skill that creates local HTML files from your provided optimization solution. If the solution data is sensitive or you need offline use, replace the CDN scripts and fonts with trusted local copies before opening the generated page.

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 (3)

Context-Inappropriate Capability

Medium
Confidence
95% confidence
Finding
The skill directs the agent to generate HTML that loads React, ReactDOM, Babel, Tailwind, and Google Fonts from third-party CDNs even though the stated task is local visualization and explicitly says no external API is needed. This creates unnecessary outbound network dependencies, leaks usage metadata to third parties, and introduces supply-chain risk if any CDN resource is unavailable or compromised.

Intent-Code Divergence

Medium
Confidence
92% confidence
Finding
The documentation claims the solution can be produced without external APIs, but the provided template still requires fetching multiple external resources at render time. This mismatch is security-relevant because it can mislead users and reviewers into believing the output is offline/local while it actually performs network requests and inherits third-party script trust.

Vague Triggers

High
Confidence
97% confidence
Finding
The trigger conditions are extremely broad and mandate use of this skill for common phrases like 'visualize this' or 'show the result,' causing the skill to activate in many unrelated optimization or coding contexts. Over-broad triggering can route users into a workflow that generates HTML/JS artifacts and external CDN dependencies without clear intent, increasing the chance of inappropriate tool use, data exposure, or unsafe handling of untrusted code snippets.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal