Security audit

厨房避坑指南

Security checks for vulnerabilities and agentic risk

Overview

This markdown-only kitchen renovation skill has no software execution risk, but it needs review because some gas and wall-remodeling advice could encourage unsafe or noncompliant work.

Review this skill before installing if users may rely on it for real renovation decisions. It should be revised to require licensed structural, gas, electrical, ventilation, and local-code review before demolition or gas-related work, and to remove any inspection workaround or concealment advice.

SkillSpector

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Description-Behavior Mismatch

Medium

Confidence: 93% confidence
Finding: The recommendation to enclose gas piping in custom cabinetry can create a hazardous condition by concealing a critical utility line that should remain inspectable and ventilated according to local code and gas-safety practice. If a leak develops, the enclosure may delay detection, impede maintenance access, and increase the risk of gas accumulation, fire, or explosion in a kitchen environment.

Vague Triggers

Medium

Confidence: 88% confidence
Finding: The skill is configured to activate for essentially any kitchen-related renovation question, which is broader than necessary for a narrowly scoped reference skill. Over-broad routing can cause the agent to apply this opinionated guidance in contexts where the user did not ask for this specific source or where other safety, legal, or specialist guidance would be more appropriate, increasing the chance of irrelevant or overly prescriptive advice.

Natural-Language Policy Violations

Medium

Confidence: 81% confidence
Finding: The skill instructs the agent to use a fixed tone and style ('不去 AI 味', specific brand/persona voice) without checking whether the user wants that communication style. While not directly dangerous like code execution, this can override user preference, reduce neutrality, and make advice sound more authoritative or one-sided than warranted in a home-renovation context.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The content gives actionable remodeling advice involving wall demolition, changing kitchen entrances, glass replacements, and gas-related sequencing, but it does not consistently foreground the need for licensed structural, gas, and code-compliance review before work begins. In a home-renovation skill, users may treat these steps as practical guidance and attempt unsafe or noncompliant modifications, creating risks of structural damage, fire/explosion, failed gas inspection, or occupant injury.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The content encourages wall demolition and layout alteration in a kitchen while presenting multiple implementation ideas in a promotional tone, but it does not prominently require structural assessment, code compliance, gas/ventilation review, or licensed professional approval before work begins. The brief closing note to confirm non-load-bearing status and local ventilation policy is insufficient relative to the repeated actionable demolition suggestions, creating a meaningful risk of unsafe structural changes, fire/code violations, and gas or ventilation hazards.

Static analysis

No suspicious patterns detected.