Back to skill

Security audit

AI机票预订助手

Security checks for vulnerabilities and agentic risk

Overview

This flight-booking skill matches its stated purpose, but it handles credentials, identity data, and refunds with unsafe transport, storage, and disclosure practices that need review before installation.

Review this skill carefully before installing. It may be useful for its intended flight-booking workflow, but users should understand that it stores an API key and phone locally, sends PII to an external flight API, prints sensitive passenger details, and submits cancellation, rebooking, or refund operations through scripts. It should be fixed to use verified TLS, private credential storage, masked PII output, safer per-session state, and explicit confirmations before transactional actions.

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 (4)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/common.py:163
Finding

TLS Certificate Verification Disabled for Sensitive API Requests

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/common.py:20
Finding

API Credentials Stored in a Predictable Shared Temporary File

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/create_order.py:90
Finding

Passenger Identity Data Exposed Through Process Arguments and Console Output

Content
View full analysis
") print("Example: python3 create_order.py 1 \"Example Name\" \"13800138000\" \"110101199001011234\"") sys.exit(1) seat_index = sys.argv[1] passenger_name = sys.argv[2] passenger_phone = sys.argv[3] passenger_id = sys.argv[4] print(f"\nCreating order...") print(f"Seat index: {seat_index}") print(f"Passenger name: {passenger_name}") print(f"Passenger phone: {passenger_phone}") print(f"Passenger identity number: {passenger_id}") ``` Order-detail output also reveals full passenger data: ```python print(f" - Name: {name}") print(f" - Phone: {phone}") print(f" - Identity-document type: {id_type_name}") print(f" - Identity-document number: {id_card}") ``` ### Technical Analysis The booking interface requires the passenger's name, phone number, and identity-document number as command-line arguments. Sensitive command-line values can be exposed through: - Process inspection tools while the command is running - Agent execution telemetry - Shell history - Debugging or monitoring systems - Audit logs that record invoked commands - Crash reports and support bundles The script then prints all three values in full. Standard output is commonly captured by the invoking Agent, terminal recording, CI logs, application logs, or conversation traces. Displaying full PII is not required to submit a booking. The Skill documentation explicitly states that PII must not be exposed in logs or responses, making the implementation inconsistent with its declared privacy behavior. ### Attack Path 1. A user supplies passenger PII to create a booking. 2. ...[truncated 1129 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/search_price.py:25
Finding

Booking and Rebooking State Trusted from Predictable Temporary Files

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 (33)

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The skill handles and displays sensitive passenger data such as phone numbers, ID numbers, and ticket numbers, but this data handling is not fully reflected in the declared purpose. In a travel-booking context, under-disclosed PII processing raises privacy and compliance risk because the skill may reveal or persist sensitive data beyond what users expect.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The skill handles and displays sensitive passenger data such as phone numbers, ID numbers, and ticket numbers, but this data handling is not fully reflected in the declared purpose. In a travel-booking context, under-disclosed PII processing raises privacy and compliance risk because the skill may reveal or persist sensitive data beyond what users expect.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The skill handles and displays sensitive passenger data such as phone numbers, ID numbers, and ticket numbers, but this data handling is not fully reflected in the declared purpose. In a travel-booking context, under-disclosed PII processing raises privacy and compliance risk because the skill may reveal or persist sensitive data beyond what users expect.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The code explicitly disables TLS certificate verification for all outbound API calls by using an unverified SSL context. In a flight-booking assistant that handles authentication and booking data, this enables man-in-the-middle interception or tampering with apiKey, phone numbers, passenger details, and booking requests, defeating transport security entirely.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
94% confidence
Finding

The skill invokes Python scripts that use network access, read/write files, and likely consume environment-provided secrets, yet it declares no explicit tool scope or permission boundaries. In a skill that handles booking, refunds, and passenger PII, this omission increases the chance of over-privileged execution, unintended data access, or exfiltration through external HTTP calls.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

L003 及全文说明均以“AI机票助手”定位并默认使用中文场景,但没有说明是否支持其他语言或允许用户选择语言。根据规则,强制特定语言而未提供用户 opt-in 或明确的地域合规理由,属于自然语言策略风险。

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger phrases for flight-related actions are broad enough to match ordinary conversation, increasing the chance the skill activates without clear user intent. Because this skill can initiate account-linked lookups and later lead into booking flows, false activations can expose travel data or start sensitive transactions in the wrong context.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The cancellation triggers include vague phrases such as '不要了' and '取消这个订单' without strong context checks. In a live booking skill, ambiguous cancellation language can cause accidental destructive actions against an order, particularly if prior context is stale or the wrong order is in scope.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The refund-apply triggers include broad phrases like '退票', which can ambiguously refer to asking about policy, fees, or actually submitting a refund. In a transactional context, such ambiguity is dangerous because a misunderstood trigger can progress a financially impactful and potentially irreversible refund workflow.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The example workflow contradicts the documented command signatures by appending an extra phone argument to several order-related scripts. Interface inconsistencies in transactional skills are dangerous because they can cause the wrong script invocation, accidental disclosure of PII on the command line, or bypass of intended validation paths.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The refund-fee text states the system will automatically submit a refund application after confirmation, which conflicts with the documented requirement for a separate explicit refund-apply step. In a refund flow, ambiguous control logic can lead to unintended irreversible actions or social-engineering-style consent where the user does not realize a second destructive step is being triggered.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The module docstring and all user-facing CLI messages are written only in Chinese, indicating the skill is designed to operate in a single language by default. Under the policy, locale or language restrictions should be optional or clearly justified; this file provides no user opt-in or explanation that the skill is region-specific.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

This code hard-codes all user-facing text in Chinese, including the usage help, progress message, and error text. That creates a natural-language policy issue if the skill is expected to be usable generally, because it does not offer user opt-in or document that it is intentionally limited to a Chinese-speaking context.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The module stores apiKey and phone number in a predictable file under the system temporary directory, which is often shared, weakly permissioned, or accessible by other local processes. This creates a local credential exposure risk and extends the lifetime of sensitive data beyond the immediate authentication session.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

Sensitive credentials and personal data are persisted locally without any visible user-facing notice or consent mechanism in this module. While this is partly a privacy/compliance issue, it also has security implications because users and operators may be unaware that reusable authentication material is left on disk.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The function automatically injects the stored apiKey and sends it, along with business parameters and optional personal fields, to a remote API endpoint. While the code comments describe the API call technically, there is no user-facing disclosure that sensitive data may be transmitted off-host.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The script sends highly sensitive personal data, including name, phone number, and government ID number, directly to an API without any visible consent prompt, privacy notice, minimization, or masking. In a flight-booking skill this data may be operationally necessary, but collecting and transmitting it without clear user-facing safeguards increases privacy, logging, and compliance risk if the API, transport, or downstream systems are mishandled.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The script prints the passenger's full phone number and full ID card number to the console, exposing sensitive PII to shell history capture, terminal logs, screen recordings, shared sessions, and monitoring systems. In a booking workflow this is especially risky because identity document numbers are high-value data that can enable identity theft or account recovery abuse if leaked.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The script prints the full endorsement application payload immediately before submission, which includes sensitive travel-order metadata such as order_id and ticket_ids and may also include additional passenger or itinerary fields from seat_item_copy. In a flight booking/change workflow, terminal logs, agent traces, or orchestration logs are often persisted or exposed to operators, so unnecessary disclosure of booking identifiers creates avoidable privacy and account-targeting risk.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

Sensitive order and ticket identifiers are emitted to output without masking or explicit user need, which can leak booking references into console history, centralized logs, screenshots, or agent telemetry. In this skill context, those identifiers are directly tied to flight change operations, so disclosure could aid unauthorized inquiry, social engineering, or misuse against customer travel records.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This Python file contains docstrings, usage text, status messages, and output labels entirely in Chinese, and there is no indication that the skill is intentionally limited to a Chinese-speaking or region-specific context. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

This Python file contains natural-language comments, docstrings, usage text, and runtime output entirely in Chinese, with no indication that the skill is region-specific or that users can choose another language. Under the policy rule for language/locale, forcing a specific language without opt-in is a natural-language policy violation.

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 full passenger personally identifiable information, including name, phone number, identity document number, and ticket number, directly to stdout with no masking, minimization, or authorization check. In a flight-booking context this data is highly sensitive; exposure through terminal history, logs, screenshots, shared shells, or agent output channels can leak customer data and facilitate privacy violations or identity fraud.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The script performs a refund submission immediately after fetching order details, with no explicit user confirmation or second-step verification before triggering a financially sensitive and potentially irreversible action. In a flight-booking/refund skill, this is more dangerous because accidental invocation, argument mix-ups, or prompt/agent mistakes can directly cause unwanted refund requests affecting real customer itineraries and payments.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

This file contains multiple natural-language strings and comments in Chinese, including docstrings, CLI usage, and runtime output. Under the policy, forcing a specific language without user opt-in is a locale/language policy violation unless the constraint is clearly documented and justified.

Content

No source excerpt is available for this finding.