Back to skill

Security audit

Rau — 前端全栈AI时代工程心智

Security checks for vulnerabilities and agentic risk

Overview

This is a Chinese-language frontend-advice persona skill with broad activation terms, but it contains no code, data access, persistence, or hidden execution.

Install this if you want a Chinese frontend-engineering review persona. Be aware that generic React, Next.js, performance, or developer-experience questions may activate it more often than expected; prefer explicit invocation if your skill manager supports 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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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 (4)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
72% confidence
Finding

The skill is entirely written in Chinese and presents its activation phrases, behavior, and outputs in that language, but it does not indicate that users may choose another language or that the Chinese-only constraint is intentional and justified. This can amount to a language/locale policy issue when users have not opted into that language.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger phrases are very broad, including common terms like React, Next.js, 性能优化, and 前端工程. In a multi-skill environment, this can cause unintended activation during ordinary frontend discussions, leading the assistant to adopt this persona when the user did not explicitly request it and potentially override more relevant instructions or context.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The activation section describes a strong persona shift and an exit word, but it does not clearly define the boundary between discussing the persona and activating it. That ambiguity can cause prompt-routing errors where the model enters persona mode based on incidental references, which may alter tone, priorities, or response policy unexpectedly.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The invocation examples are open-ended and map to extremely common frontend requests such as React component design, performance optimization, and architecture choices. Without constraints, these examples effectively broaden activation to routine engineering conversations, increasing the chance of accidental skill takeover and reducing predictability of assistant behavior.

Content

No source excerpt is available for this finding.