ClawHub

Huo15 Openclaw Design Critique

Security checks across malware telemetry and agentic risk

Overview

This is a design-review prompt skill with broad trigger wording, but its behavior is disclosed, purpose-aligned, and does not install code or request privileged access.

Install if you want a structured, opinionated UI critique workflow. Be aware that its broad triggers may make it activate on general design-feedback requests, and avoid sharing screenshots or URLs that expose private data unless you intend the assistant to analyze them.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Vague Triggers

Medium
Confidence
89% confidence
Finding
The alias and trigger set is very broad, including generic phrases like '设计评审', 'UI 审查', and 'design review'. In a multi-skill agent, this can cause accidental or overly eager activation on ordinary design-feedback requests, routing users into this rigid workflow when they may have wanted general assistance instead.

Vague Triggers

Medium
Confidence
84% confidence
Finding
The activation section mixes broad examples of when to run with refusal conditions later in the workflow, but does not define a clear boundary between triggering and declining. That ambiguity can lead the agent to invoke the skill prematurely, then force users into unnecessary screenshot/URL handling steps or refuse after partial engagement.

Vague Triggers

Medium
Confidence
90% confidence
Finding
The standalone trigger list contains short, generic phrases such as '优化建议', '改进建议', and '这个页面有什么问题', which overlap heavily with normal conversation. Without constraints, these phrases can hijack unrelated requests and increase false activations across the assistant.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal