Back to skill

Security audit

智慧旅行预订

Security checks for vulnerabilities and agentic risk

Overview

This is an incomplete hotel-search aggregation prototype with disclosed multi-platform API intent and no evidence of hidden persistence, credential theft, destructive actions, or implemented booking/payment flows.

Before installing, treat this as a prototype rather than a working booking assistant. If real API integrations are added, confirm that hotel search details may be sent to the named providers and review any credentials, booking, payment, or order-submission logic separately.

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
  • 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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (8)

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

描述与代码方向相关,都是酒店聚合场景,但存在实质性能力落差。代码没有真正调用分贝通、携程、美团、同程、华住会、锦江等外部接口,六个搜索函数全部是 TODO 并直接返回空结果,因此“整合多个酒店数据源”目前并未实际实现。其次,描述中的“房型查询、预订服务”在代码中也没有对应完整能力:只有 RoomSource / AggregatedRoom 数据结构,没有房型查询API;也没有任何预订相关方法、订单创建、鉴权、支付或提交动作。现有代码更准确地说是一个酒店聚合搜索框架/原型,而不是描述中的完整聚合与预订助手。因此应判定为描述与实际行为不一致。

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
93% confidence
Finding

The skill clearly describes mandatory calls to multiple external platform APIs, but it does not declare any explicit tool scope, permissions, or allowed-tools boundaries. This creates an authorization and governance gap: an agent may be permitted to make network calls without transparent limitation, making unintended outbound requests or access to additional endpoints more likely.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill promises booking and aggregation across several external platforms but provides no warning that user-supplied travel data may be shared with those third parties. In a travel context, this can expose location, travel dates, corporate travel patterns, and potentially identity-linked booking preferences, creating privacy and compliance risk.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The instruction that the skill 'must call each platform API to obtain real data' encourages external transmission by default, yet no user-facing warning or consent mechanism accompanies that requirement. Because hotel searches inherently include itinerary details and sometimes employer-linked travel intent, silent API calls to multiple vendors create meaningful privacy and data-sharing exposure.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger condition is broad enough to activate on ordinary hotel-search requests without clearly signaling that the skill may query multiple third-party services. In this context, overbroad triggering increases the chance that user travel queries, dates, destinations, and related preferences are sent externally without sufficiently specific user intent or consent.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

SQP-3 applies to all file types and includes language/locale policy violations in natural-language comments or strings. The file contains operational instructions entirely in Chinese and mandates behavior based on them, without any user opt-in or alternative language support.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
94% confidence
Finding

The formatted output tells users to reply with a hotel index to view room details or with "序号-比价" to compare platform prices, but this file defines no handler or method that implements either interaction. This is an active documentation/UI-intent mismatch rather than mere incompleteness, because the code explicitly instructs the user to invoke capabilities that are absent.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
90% confidence
Finding

The comments state '如非此版本,请拒绝执行' ('if not this version, refuse to execute'), and the file consistently hardcodes Chinese-language descriptions and messages. For a general adapter, this creates a locale/policy constraint in natural language without offering user choice or documenting a justified regional limitation.

Content

No source excerpt is available for this finding.