Back to skill

Security audit

AI机票预订助手

Security checks for vulnerabilities and agentic risk

Overview

The skill appears to be a real flight-booking helper, but it handles credentials, identity data, and ticket transactions with unsafe storage, logging, and network protections.

Review before installing or using for real bookings. Only proceed if you trust the external flight service and can tolerate the current risks; the safer path is to enable normal TLS verification, restrict the API endpoint, move credentials and transaction files into user-private storage with restrictive permissions, redact passenger identity data from output, and require an explicit separate confirmation before refund or change submission.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (5)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/common.py:157
Finding

TLS Certificate Verification Is Disabled for All API Requests

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/common.py:15
Finding

Environment-Controlled API Endpoint Can Redirect Authentication Secrets and Passenger Data

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/common.py:20
Finding

Long-Lived API Key Is Stored in a Predictable Shared Temporary File

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/create_order.py:96
Finding

Full Passenger Identity Data Is Exposed in Command Output

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/search_price.py:26
Finding

Predictable Temporary Transaction Files Can Be Tampered With Before Booking or Endorsement

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (32)

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The declared description focuses on flight search and ticket operations, but the skill also performs phone-based authentication, verification-code handling, apiKey acquisition, and persistent credential storage. Undisclosed auth and credential-management behavior is security-relevant because it changes the trust boundary and causes users to provide sensitive data and create persistent local state they were not clearly warned about in the top-level declaration.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The declared description focuses on flight search and ticket operations, but the skill also performs phone-based authentication, verification-code handling, apiKey acquisition, and persistent credential storage. Undisclosed auth and credential-management behavior is security-relevant because it changes the trust boundary and causes users to provide sensitive data and create persistent local state they were not clearly warned about in the top-level declaration.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The declared description focuses on flight search and ticket operations, but the skill also performs phone-based authentication, verification-code handling, apiKey acquisition, and persistent credential storage. Undisclosed auth and credential-management behavior is security-relevant because it changes the trust boundary and causes users to provide sensitive data and create persistent local state they were not clearly warned about in the top-level declaration.

Content

No source excerpt is available for this finding.

Vague Triggers

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger phrase '退票' is overly broad for a destructive refund-submission action and could match users merely asking about refund policy, eligibility, or fees. In this context, where refunds affect paid orders and may be hard to reverse, over-triggering materially raises the chance of unintended transaction execution.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The refund display text says the system will automatically submit a refund application, which directly conflicts with the stated requirement to wait for explicit user confirmation before performing the destructive action. In a financial/transactional workflow, this ambiguity can cause unintended refunds or unauthorized irreversible actions if an agent follows the wrong instruction path.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The code disables TLS certificate verification by creating an unverified SSL context for outbound API calls. This allows man-in-the-middle attackers to intercept or modify flight search, booking, authentication, and personal data traffic, which is especially dangerous in a travel-booking skill handling credentials and user PII.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding

The skill declares no explicit tool scope or permissions despite instructing use of Python scripts that perform network access, local file reads/writes, and access to auth state in the user's home directory and /tmp. This weakens containment and reviewability, making it easier for the skill to exercise broader capabilities than a host or user may expect, especially when handling PII and persistent credentials.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The natural-language description and all invocation guidance are presented only in Chinese, and the skill name/behavior do not state that it is intentionally limited to Chinese-speaking or region-specific users. Under the stated policy, forcing a specific language without opt-in is a locale-policy concern.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

Using the broad trigger phrase '预订' can activate the skill for loosely related travel-booking requests and may push the agent into a transactional flow prematurely. Because this skill can lead to PII collection, external API calls, and order creation, overbroad activation increases the risk of unintended data sharing or accidental commerce actions in ambiguous conversations.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The full-flow examples contradict the earlier command definitions by adding phone-number parameters to later order operations that were not specified in the canonical command interface. In security-sensitive transaction flows, contradictory invocation semantics can lead to wrong parameter binding, accidental disclosure of phone numbers on the command line, bypass of expected checks, or implementation drift where agents invoke the wrong action with sensitive identifiers.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The module docstring and all user-facing CLI messages are written only in Chinese, indicating a fixed language choice for the skill. The file does not offer any language/locale opt-in or explain that the skill is intended only for a Chinese-speaking or region-specific context.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest describes flight search, cabin/price queries, booking, ticket changes, and ticket refunds, but does not mention order cancellation. This script directly calls a cancelOrder operation, which is a distinct order-management capability not clearly covered by the declared description.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill stores the apiKey, phone number, and authentication metadata in a predictable file under the system temporary directory. Temporary directories are often accessible to other local users or processes and are not appropriate for long-lived secrets, creating risk of credential theft or session hijacking.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The comment claims certificate verification should only be bypassed outside production, but the live implementation actively disables verification. This mismatch increases operational risk because reviewers or operators may wrongly assume TLS is being validated when it is not.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The function automatically injects the apiKey and sends it along with business parameters and extra personal fields such as phone or idCard over the network. While the docstring describes the API call for developers, there is no user-facing disclosure that potentially sensitive personal and authentication data will be transmitted to a remote service.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The authenticated API path repeats the same insecure TLS-bypass pattern while suggesting verified TLS should be used. Because this path sends apiKey and booking-related data, the contradiction can hide a real transport-security failure in the most sensitive request flow.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The script collects and transmits highly sensitive personal data, including name, phone number, and government ID number, directly to an API and also prints that data to stdout. In a flight-booking context this data flow is expected for legitimate order creation, but the lack of masking, privacy notice, minimization, or safeguards increases the risk of PII exposure through logs, terminals, shell history, or downstream API handling.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The script prints the full order_data object immediately before submitting the endorsement request, which likely includes order identifiers, ticket/product IDs, and flight/segment details. In a flight-booking skill, these are sensitive transactional data and may be exposed in terminal logs, agent traces, or shared session output, enabling unauthorized viewing or reuse of booking-related information.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

This code uses Chinese-only docstrings, status messages, errors, and usage examples throughout the script, which enforces a specific language for users. The file does not offer any language selection, fallback, or justification that the skill is intentionally limited to a Chinese-only regional context.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

Docstrings, usage text, status messages, and prompts are all presented only in Chinese. This forces a specific language/locale without any opt-in, fallback, or documented region-specific constraint, which matches the stated natural-language policy concern.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The script writes order- and flight-related data, including seat options and booking context, to a temporary JSON file on disk without any visible access controls, retention policy, or user warning. In a flight-booking context this data can be sensitive operational or personal travel information, and storing it in temp locations increases the risk of unintended disclosure to other local users, processes, or later runs.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The script performs a state-changing refundApply API call immediately after gathering product IDs, without any explicit user confirmation, review step, or safeguard against accidental execution. In a flight-booking skill, refunds are financially sensitive and can be difficult or impossible to reverse once submitted, so accidental or induced invocation can cause real monetary and operational harm.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The file contains comments, usage text, status messages, and output headers exclusively in Chinese, including user-facing strings at L32, L40, L60, and L65-L76. This imposes a specific language on all users without any opt-in, configuration, or documented justification for a locale restriction.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

This file contains natural-language instructions, errors, and status messages entirely in Chinese, including the module docstring and CLI output. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale restriction is explicitly documented and justified, which is not present here.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

This code uses Chinese-only docstrings, comments, usage text, and runtime output throughout the file, which imposes a specific language/locale on users without any opt-in or alternative. The policy explicitly flags language or locale restrictions when the skill does not offer user choice or document a justified regional constraint.

Content

No source excerpt is available for this finding.