Back to skill

Security audit

AI机票预订助手

Security checks for vulnerabilities and agentic risk

Overview

This flight-booking skill is mostly purpose-aligned, but it handles credentials, personal data, payments, changes, and refunds with unsafe storage, weak network protections, and unclear confirmation boundaries.

Review carefully before installing. Use only in a trusted local environment, avoid shared machines, assume terminal output and process arguments may expose passenger identity details, and do not use it for real bookings, changes, or refunds until TLS verification, credential storage, endpoint validation, redaction, and explicit confirmation safeguards are fixed.

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 Validation 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 Exfiltrate Credentials and Passenger Data

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/common.py:20
Finding

API Key and Phone Number Are Stored in an Unsafe Predictable Temporary File

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/create_order.py:89
Finding

Verification Codes and Passenger PII Are Exposed Through Process Arguments and Console Output

Content
View full analysis
") print("示例: python3 create_order.py 1 \"张三\" \"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"\n正在创建订单...") print(f"舱位编号: {seat_index}") print(f"乘客姓名: {passenger_name}") print(f"乘客手机号: {passenger_phone}") print(f"乘客证件号: {passenger_id}") ``` Authentication also accepts the SMS code through a command-line argument and prints the phone number: ```python phone = sys.argv[2] code = sys.argv[3] verify_and_get_api_key(phone, code) ``` Order-detail output prints returned PII without masking: ```python if name: print(f" - 姓名: {name}") if phone: print(f" - 联系电话: {phone}") if id_card and id_type_name: print(f" - 证件类型: {id_type_name}") print(f" - 证件号: {id_card}") if ticket_no: print(f" - 票号: {ticket_no}") ``` ### Technical Analysis Sensitive values are supplied as command-line arguments. On many operating systems, command lines can be observed through process-inspection facilities while a process is running and may be retained in shell history, orchestration logs, Agent tool-call records, or diagnostic telemetry. The booking script then prints the full passenger name, phone number, and identity-document number. The order-detail script prints the same data returned by the server, along with ticket numbers. This directly contradicts the Skill documentation, which says not to expose personal information ...[truncated 1476 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/search_price.py:24
Finding

Predictable Temporary Booking-State Files Can Be Tampered With Before Transaction Submission

Content
View full analysis
Remediation
View remediation
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 (28)

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The declared description focuses on flight assistance, but the skill also performs authentication flows involving SMS verification, apiKey issuance, and local credential persistence. Hidden or under-disclosed auth and secret-handling behavior reduces transparency and can mislead operators about what sensitive actions and data storage the skill performs.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The declared description focuses on flight assistance, but the skill also performs authentication flows involving SMS verification, apiKey issuance, and local credential persistence. Hidden or under-disclosed auth and secret-handling behavior reduces transparency and can mislead operators about what sensitive actions and data storage the skill performs.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The declared description focuses on flight assistance, but the skill also performs authentication flows involving SMS verification, apiKey issuance, and local credential persistence. Hidden or under-disclosed auth and secret-handling behavior reduces transparency and can mislead operators about what sensitive actions and data storage the skill performs.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The code explicitly disables TLS certificate verification for outbound API requests by using an unverified SSL context. This allows a man-in-the-middle attacker to intercept or modify flight search, booking, refund, and authentication traffic, including apiKey and passenger-related data, which is especially dangerous given this skill handles travel transactions.

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 can access the network, read/write local files, and potentially inspect environment data, yet it declares no explicit tool scope or permissions boundary. This creates an over-privileged integration surface where a host agent may allow capabilities beyond what users and reviewers expect, increasing the chance of data exposure or unintended side effects.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The manifest description and the rest of the markdown are written as a Chinese-only interaction pattern, and no language-choice or opt-in behavior is described. Under the stated policy, forcing a specific language without user choice is a natural-language policy concern unless the locale restriction is clearly documented and justified.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

Using the broad trigger phrase '预订' can cause the skill to activate on casual planning or informational requests that do not indicate consent to place an order. In a booking context tied to PII collection and external service calls, ambiguous invocation increases the chance of premature transactional behavior.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

Using the broad trigger phrase '退票' for a destructive flow is dangerous because users may be asking about rules or fees rather than authorizing cancellation/refund processing. In this skill, refund actions affect existing paid travel and can be irreversible or costly, so ambiguous intent handling materially raises the risk of unauthorized destructive operations.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The full-process examples use command signatures that differ from the earlier defined interfaces, including extra phone parameters in several order-related operations. In security-sensitive flows, signature inconsistencies can cause the agent to pass the wrong data, mishandle identity binding, or invoke scripts in unintended ways, increasing risk of erroneous destructive actions or PII leakage.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill states that refund submission must require explicit user confirmation, but the sample output says the system will automatically submit the refund request after agreement. Contradictory instructions in a destructive flow can lead an agent to take irreversible refund actions without a clear confirmation boundary.

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 is entirely in Chinese and the CLI messages throughout the file are presented only in Chinese, indicating a fixed language experience. There is no indication that users can opt into another language or that the skill is intentionally restricted to a Chinese-only regional context, which matches the language/locale policy violation criteria.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill stores the apiKey and phone number in a predictable file under the system temporary directory, which is often accessible to other local users, processes, or cleanup tools depending on platform configuration. Because the file contains reusable authentication material and PII, local disclosure could enable unauthorized API use or account misuse.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
85% confidence
Finding

The function sends the stored API key and business parameters over the network to a remote API endpoint. While this is part of the helper's function, the code itself provides no user-facing disclosure that authentication data and request contents are transmitted to an external service.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The script prints the passenger's full name, phone number, and government ID number directly to stdout before creating the order. In CLI, agent, or logged execution environments, stdout is often captured in terminal history, orchestration logs, transcripts, or monitoring systems, which can expose personally identifiable information and increase privacy, compliance, and identity-theft risk.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

Natural-language strings throughout the script, including docstrings, status messages, errors, and usage text, are written only in Chinese. This imposes a single language on all users with no opt-in, fallback, or explanation that the tool is intended exclusively for a Chinese-speaking or region-specific environment.

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 endorsement request payload before calling the API, which can expose sensitive booking data such as order IDs, ticket/product identifiers, and detailed segment information in terminal logs, orchestration logs, or shared execution environments. In a flight-booking skill, these identifiers are tied to real customer travel operations, so leaking them increases privacy and account abuse risk even if they are not direct authentication secrets.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

This Python file contains natural-language comments, usage text, status messages, and output labels entirely in Chinese. Because the skill does not offer any user opt-in or alternative locale, it appears to enforce a specific language, which matches the policy-violation category for language or locale constraints.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

This code contains natural-language comments, help text, status messages, and prompts only in Chinese, such as the usage output and interactive instructions. Under the policy, forcing a specific language without offering a user language/locale choice can be a natural-language policy violation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The script retrieves and prints highly sensitive booking data including passenger names, phone numbers, identity document numbers, and ticket numbers directly to stdout with no masking, consent check, or warning. In an agent skill context, stdout may be surfaced to users, logs, transcripts, or operator tooling, so exposing full PII can cause privacy violations and unauthorized disclosure even if the underlying API call is legitimate.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The script submits a refund request immediately after fetching order details, with no confirmation prompt, no explicit warning that the action is financially impactful, and no validation that the refund amount or passenger selection is intended. In a flight-booking skill, refunds are operationally significant and may be irreversible or costly once sent to the backend, so accidental execution can directly affect customer bookings and finances.

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 status output entirely in Chinese, including the command help and result messages. Under the policy for natural-language violations, forcing a specific language without user opt-in is a locale/language policy issue unless the tool is clearly documented as region-specific, which is not shown here.

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 messages in the script are in Chinese, and the skill provides no option for users to select another language or locale. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless the regional constraint is explicitly justified.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill description advertises a wide range of flight-related actions, including booking, rescheduling, and refunds, but does not define explicit trigger conditions, scope boundaries, or user-intent constraints. In an agent ecosystem, broad activation language can cause the skill to be invoked for loosely related travel conversations, increasing the chance of unintended transactional behavior or over-collection of user data.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
96% confidence
Finding

The script's user-facing strings are written entirely in Chinese, including error text, usage instructions, and status messages. Because the file provides no user opt-in or configurable locale selection, it appears to enforce a specific language, which matches the language/locale policy violation category.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
72% confidence
Finding

The module description and all user-facing strings are written only in Chinese, with no indication that the skill is region-specific or that other language options are supported. Under the stated policy, forcing a specific language without opt-in can be a locale-policy issue.

Content

No source excerpt is available for this finding.