Back to skill

Security audit

zhenxi-audio

Security checks for vulnerabilities and agentic risk

Overview

The audio workflow is coherent, but it permits adding your registration email to a separate donation site's link without first asking about that sharing.

Use the donation link without the email parameter unless you explicitly agree to share your address with that site. Review configuration replacements, protect the service token, and check credit costs before purchasing; the skill states that credits are non-refundable.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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)

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · player-template.html (reported line 2)May include surrounding context.

html
<!DOCTYPE html>
<!--
  珍惜音频资源服务 · 试听播放页模板(智能体交付用)
  用法:替换全部 {{大写占位符}} 后整页交付(可渲染 artifact 的直接渲染,否则另存为 .html 让用户打开)。
    {{BOOK_TITLE}}    书名,如 基金经理

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · player-template.html (reported line 2)May include surrounding context.

html
<!DOCTYPE html>
<!--
  珍惜音频资源服务 · 试听播放页模板(智能体交付用)
  用法:替换全部 {{大写占位符}} 后整页交付(可渲染 artifact 的直接渲染,否则另存为 .html 让用户打开)。
    {{BOOK_TITLE}}    书名,如 基金经理

The donation-link workflow includes the user's registration email in a third-party URL without a privacy warning.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

The workflow explicitly permits placing the user's registration email in a donation URL on a different domain without explaining the disclosure or requesting consent. Opening that link sends the email to the donation service and can expose it through browser history and URL logging; its account-matching purpose does not eliminate the privacy risk.

Content

No source excerpt is available for this finding.