Back to skill

Security audit

Tts Cleanup Checklist

Security checks for vulnerabilities and agentic risk

Overview

This is a text-cleanup skill for preparing files for TTS, with some usability caveats but no hidden execution, credential access, persistence, or exfiltration behavior.

Installers should be aware that this skill may remove citation markers, bare URLs, and reference sections by default, and its report labels are Chinese. For source-sensitive documents, ask the agent to preserve citations/links and set the desired output language before running it.

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
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Natural-Language Policy Violations

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill metadata and description are written to steer use toward Chinese-language behavior without giving the user a language-choice mechanism. This can override user expectations and produce outputs or labels in a language the user did not request, reducing transparency and potentially impairing review of modifications.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The fixed output template mandates Chinese headings and labels such as '清洗后文本' and '清洗说明', with no opt-in or language negotiation. While not a code-execution issue, this is a real policy and usability weakness because it can force unreadable or unexpected output formats and obscure what transformations were applied.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill description includes broad trigger phrases like '只要用户提到…就应优先调用本技能', which can cause the agent to invoke this skill on loosely related requests without sufficient intent confirmation. That creates prompt-routing risk: the agent may transform or strip content when the user did not explicitly ask for TTS cleanup, leading to unintended data modification or workflow hijacking.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The instruction '当用户要求“清洗用于朗读”时,Agent 直接按本技能执行' is underspecified because '用于朗读' can overlap with many editing, summarization, accessibility, or formatting tasks. Direct execution without validating scope or user preferences increases the chance of silent content alteration, especially because the skill removes references, links, and structural material by default.

Content

No source excerpt is available for this finding.