Security audit

复式·跃层·LOFT装修指南

Security checks for vulnerabilities and agentic risk

Overview

This is a markdown-only renovation advice skill with some leftover promotional and biased reference text, but no code, persistence, credential access, or high-impact agent authority.

Before installing, be aware that the skill is a renovation reference corpus and may repeat promotional remnants or dated style assumptions from source articles. Verify structural, electrical, waterproofing, fire-safety, and property-rule decisions with qualified local professionals before acting on the advice.

A.I.G

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

SkillSpector

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 (3)

Description-Behavior Mismatch

Medium

Confidence: 93% confidence
Finding: The reference content includes product purchase prompts and teaser-style promotional material that goes beyond neutral装修 guidance. In a skill context, this can steer users toward commercial content, reduce trust in the knowledge base, and create risk of undisclosed advertising or link injection if surfaced directly to users.

Natural-Language Policy Violations

Low

Confidence: 95% confidence
Finding: The content repeatedly frames the occupant and user needs as specifically female (for example, '美女' and references to beauty-related preferences) without any user choice or necessity tied to the renovation guidance. This can create exclusionary or biased behavior in the skill’s responses, reducing suitability for broader users and potentially reinforcing stereotypes, though it is not a traditional security exploit.

Vague Triggers

Medium

Confidence: 87% confidence
Finding: The keyword recommendation block contains broad, clickable trigger phrases such as '免费设计', '装修合同错', and '善良却被坑' without any explicit rule limiting when they should be surfaced. In an agent skill, this can cause over-triggering, topic hijacking, or unintended navigation into unrelated or manipulative content paths, especially if the runtime treats these as activation cues.

Static analysis

No suspicious patterns detected.