Back to skill

Security audit

便捷旅行预订

Security checks for vulnerabilities and agentic risk

Overview

The skill is a hotel-search skeleton that overstates live search and booking abilities and lacks clear privacy/consent limits for multi-platform travel queries.

Review before installing. This skill should not be relied on for real hotel availability, pricing, room details, or booking until the provider integrations and consent/privacy controls are implemented. Require explicit user approval before sending travel search details to third-party platforms.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (8)

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

描述强调这是一个可用的酒店聚合助手,能够整合多个指定平台并提供搜索、房型查询、预订服务。但代码中的核心平台搜索方法(分贝通、携程、美团、同程、华住会、锦江)全部是 TODO 占位,实际不访问任何数据源,聚合搜索最终只会在空结果上运行。代码确实包含与酒店聚合相关的支持性结构和逻辑,例如并发调度、去重、最优价格计算、缓存和格式化展示,因此主题方向与描述相关;但其实际可执行能力远弱于声明,尤其缺失真实多源搜索、房型查询接口和预订能力,属于功能性描述与实现明显不符。

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding

The skill explicitly requires calling multiple external platform APIs to fetch hotel data, but it declares no tool scope or permission boundaries for network access. This creates an authorization and governance gap: an agent may invoke outbound requests without explicit user-visible constraints, making unintended data sharing or overbroad external access more likely.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The trigger scope is broad enough that the skill may activate for generic hotel-related requests without clear limits or user confirmation. In a multi-platform aggregation context, that can cause unnecessary external lookups, unintended sharing of travel parameters, or execution in situations where the user did not ask for cross-platform querying.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill says it will call multiple hotel platform APIs but provides no privacy warning about transmitting user search details to third parties. Travel dates, destinations, and potentially corporate booking context can be sensitive, so silent multi-party sharing exposes users to privacy, policy, and compliance risks.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The activation rule 'when the user searches hotels, query multiple platforms' lacks constraints about when aggregation is appropriate, what data is shared, and whether the user has opted in. Because hotel searches commonly include sensitive itinerary details, automatic multi-platform querying increases privacy and compliance risk.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
81% confidence
Finding

This code includes a natural-language directive, '如非此版本,请拒绝执行', alongside other user-facing comments in Chinese. Because the file does not offer any language or locale choice and the instruction is tied to a single language context, it creates a locale-specific constraint without opt-in or justification.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill tells users they can reply with a hotel index to view room details or cross-platform price comparisons, but no corresponding implementation exists in this file. In an agent setting, this kind of overclaimed capability can mislead downstream orchestration or users into assuming unsupported booking/comparison actions are available, creating integrity and trust issues and potentially causing incorrect transactional flows.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The natural-language comments state a fixed version requirement and instruct refusal to execute otherwise, and the file consistently presents the interface text in Chinese without offering any user language choice. This can violate the language/locale policy because the skill appears to force a specific language context without opt-in or a documented regional justification.

Content

No source excerpt is available for this finding.