Back to skill

Security audit

Quality Boost - 大模型回答质量提升器

Security checks for vulnerabilities and agentic risk

Overview

This is a Chinese-language prompt-quality guide with disclosed templates and optional global configuration, not a data-accessing or executable skill.

Install this if you want Chinese-language prompt-quality templates and are comfortable with optional auto-apply/system-prompt settings changing the assistant's style across conversations. Review or avoid the global config snippets if you only want to apply the rules manually.

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
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
  • 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 (5)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The skill is written and framed in Chinese and presents Chinese-language templates and examples without any opt-in or fallback behavior, which can force a locale mismatch for users operating in another language. While not directly dangerous in a classic exploit sense, this can degrade usability, cause misunderstanding of constraints, and unexpectedly change output language or style.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill description says it offers quantitative metrics and professional evaluation, and the later reporting template presents accuracy as a scored dimension. However, in the provided JavaScript evaluator, accuracy is initialized to 0 and never computed before being included in the overall score, so the documented intent of meaningful quality assessment diverges from what the code actually does.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The trigger phrases are broad enough to activate during normal user conversation, which can cause the skill to apply unexpectedly and alter assistant behavior without clear user intent. In this skill, that means inserting extra process constraints like forced restatement, self-checking, and formatting rules that may interfere with other tasks or higher-priority instructions.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill hard-codes Chinese-only behavioral instructions and output conventions without any language negotiation, fallback, or documented scope restriction. When injected as a global system prompt, this can override user language preferences, reduce transparency, and cause unsafe misunderstandings in multilingual contexts, especially if users cannot reliably interpret constraints or safety-relevant content.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

This markdown skill content is written exclusively in Chinese, including all templates and instructions. Under the policy rule for language/locale, forcing a specific language without offering user choice can be a natural-language policy violation.

Content

No source excerpt is available for this finding.