ClawHub
Back to skill

Security audit

Webflow 中国加速

Security checks across malware telemetry and agentic risk

Overview

This appears to be a real Webflow China acceleration skill, but the included Worker should be reviewed because it can fetch and persistently cache overly broad content and silently changes site branding and SEO-related files.

Review and edit the Worker before deployment. Add a strict allowlist for Webflow asset hosts in /_cdn/, remove the Webflow badge-hiding rules unless the site owner has explicit rights to do so, decide explicitly whether robots.txt and sitemap.xml should come from R2, use a dedicated R2 bucket, and document how to purge cached assets.

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)

Tp4

High
Category
MCP Tool Poisoning
Confidence
82% confidence
Finding
The documented behavior includes actions beyond the stated acceleration purpose, notably removing the Webflow badge, overriding SEO files from R2, and deleting Set-Cookie headers to force cacheability. These side effects can alter site attribution/compliance, search-engine behavior, and session or personalization semantics without clear disclosure, making the skill more dangerous because it is presented as a narrowly scoped performance optimization tool.

Context-Inappropriate Capability

Medium
Confidence
94% confidence
Finding
The worker intentionally removes the Webflow badge and injects CSS to hide it, which is unrelated to performance optimization and alters licensing/branding behavior. In the context of a China-mainland acceleration skill, this extra capability is suspicious because it expands the worker beyond stated scope and can facilitate policy or terms-of-service circumvention.

Vague Triggers

Medium
Confidence
84% confidence
Finding
The skill mandates invocation for a very wide set of loosely related topics, which can cause the agent to over-apply this skill in contexts where it is not appropriate. Over-broad routing increases the chance of unnecessary infrastructure, DNS, caching, proxying, or compliance advice being injected into unrelated conversations, potentially leading to unsafe or irrelevant operational changes.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.