ClawHub
Back to skill

Security audit

premium-html-studio

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent HTML documentation template pack, with manageable cautions around external web assets and optional active JavaScript in generated pages.

Reasonable to install if you want richer HTML document generation. Treat generated HTML as active browser content: avoid opening documents built from untrusted text without sanitizing, and replace remote fonts/CDNs with local assets when privacy, offline use, or strict supply-chain control matters.

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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
Findings (4)

Description-Behavior Mismatch

Medium
Confidence
94% confidence
Finding
The stylesheet imports fonts from Google Fonts via an external URL, which means generated documents are not truly self-contained and will make network requests when opened. This leaks reader metadata such as IP address, user agent, and access timing to a third party, can break in offline or restricted environments, and creates a supply-chain/dependency risk if the remote asset changes or becomes unavailable.

Context-Inappropriate Capability

Medium
Confidence
91% confidence
Finding
The SVG template imports fonts from Google via `@import`, which causes network access during rendering and creates an external dependency inside what is presented as a reusable local component template. This can leak document access metadata, break offline rendering, and introduce supply-chain/privacy concerns if the remote resource changes or is blocked.

Vague Triggers

Medium
Confidence
90% confidence
Finding
The trigger list is extremely broad, including generic phrases like requests to make a professional HTML doc or proposal. Overbroad activation can cause the skill to take over unrelated conversations, steering the agent into emitting large HTML/JS/CSS payloads unexpectedly and increasing the chance that unsafe template code or injected content is propagated.

Unvalidated Output Injection

High
Category
Output Handling
Content
}
  
  const results = index.search(query);
  searchResults.innerHTML = results.map(r => {
    const section = document.querySelectorAll('section')[r.id];
    const title = section.querySelector('h2').textContent;
    return `<div class="search-result-item" onclick="scrollToSection(${r.id})">
Confidence
98% confidence
Finding
The sample search feature builds HTML with `innerHTML` using document-derived content and inline `onclick`, which is a classic client-side injection sink. If any indexed section content or heading includes attacker-controlled HTML-like payloads, generated documents can become XSS-capable, enabling script execution in the viewer's browser.

Static analysis

No suspicious patterns detected.