Back to skill

Security audit

多目的地航线组合比价助手

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent flight-comparison assistant, but it repeatedly instructs agents to bypass HTTPS certificate checks and uses loosely scoped command and HTML generation patterns that need review before installation.

Review or modify this skill before installing. Do not run the FlyAI commands with TLS verification disabled; require normal certificate validation, validate and pass command arguments safely, escape generated HTML, allow only trusted HTTPS booking URLs, and confirm whether you want travel details stored in Memory or written to shareable HTML files.

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 (3)

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:179
Finding

TLS Certificate Verification Is Globally Disabled for Flight Searches

Content
View full analysis
⚠️ FlyAI commands must use the `NODE_TLS_REJECT_UNAUTHORIZED=0` prefix. ``` ```markdown | SSL certificate verification failure | Ensure the command uses the `NODE_TLS_REJECT_UNAUTHORIZED=0` prefix | ``` ### Technical Analysis Setting `NODE_TLS_REJECT_UNAUTHORIZED=0` disables TLS certificate verification for the Node.js process running the FlyAI command. The client will therefore accept expired, self-signed, untrusted, or hostname-mismatched certificates. This removes the server-authentication guarantee normally provided by HTTPS. Encryption alone does not protect the request if the client cannot verify which server it is communicating with. A network-positioned attacker can terminate the TLS connection using an arbitrary certificate and proxy or modify the traffic. The weakness is particularly significant because FlyAI responses contain flight data and `jumpUrl` booking links that are later presented to users and embedded in generated HTML reports. ### Attack Path 1. The Skill starts a FlyAI search with `NODE_TLS_REJECT_UNAUTHORIZED=0`. 2. An attacker gains a network interception position, such as through a hostile Wi-Fi access point, compromised proxy, DNS manipulation, or local network gateway. 3. The attacker presents an untrusted TLS certificate while impersonating the FlyAI endpoint ...[truncated 1073 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:179
Finding

User-Controlled Travel Parameters Are Interpolated into Shell Commands

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
reference/html-template.md:484
Finding

Generated HTML Embeds Untrusted Text and Booking URLs Without Contextual Validation

Content
View full analysis
✈️ 预订去程 ✈️ 预订回程 ``` The same behavior is repeated for the other recommendations: ```html ✈️ 预订去程 ✈️ 预订回程 ``` ```html ✈️ 预订去程 ✈️ 预订回程 ``` The generation instructions require direct placeholder replacement without specifying escaping or URL validation: ```markdown 1. Complete all route searches and scoring. 2. Replace the `{{variable}}` placeholders in the template. 3. Use the `create_file` tool to generate the file. ``` Numerous other placeholders derived from user input or external responses are inserted directly into element text, including city names, airline names, recommendation reasons, route descriptions, visa notices, and weather notices. ### Technical Analysis The report-generation process treats user and API values as trusted HTML. Direct string replacement is unsafe because different insertion contexts require different handling: - Text inserted between tags must be HTML-escaped. - Values inserted into attributes must be attribute-escaped. - URL attributes must also undergo scheme and destination validation. A text payload containing markup can terminate the intended element and inject new HTML. ...[truncated 2125 chars]
Remediation
View remediation
Book flight ``` 10. Consider applying a restrictive Content Security Policy, such as disallowing scripts and external resource loading. 11. Add tests using malicious text and URL payloads to verify that generated reports contain encoded text rather than executable markup. 12. Keep report content and link destinations visibly identifiable so users can verify the booking provider before leaving the report. ]]>
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (41)

Missing User Warnings

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The document instructs execution of network commands with TLS verification disabled and provides no safety warning beyond operational guidance. This normalizes insecure transport as part of the expected workflow, increasing the chance that implementers will deploy a man-in-the-middle-prone design for live travel queries.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The documentation globally instructs users to run all FlyAI commands with NODE_TLS_REJECT_UNAUTHORIZED=0, which disables TLS certificate validation for Node-based HTTPS connections. This enables man-in-the-middle interception or tampering of flight-search traffic and any returned booking URLs/data, and the capability is not justified by a flight comparison skill’s purpose.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The file tells users to disable TLS certificate verification without explaining the security consequences. Because this is presented as a blanket prerequisite for normal use, users are pushed into unsafe operation where HTTPS authenticity checks are bypassed, exposing searches and results to interception and manipulation.

Content

No source excerpt is available for this finding.

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · reference/html-template.md (reported line 401)May include surrounding context.

md
</head>
<body>
  <div class="container">
    <!-- 头部 -->
    <div class="header">
      <h1>✈️ 航线比价方案</h1>
      <div class="meta">{{出发城市}} ↔ {{区域}}多城 · {{日期区间}}</div>

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · reference/html-template.md (reported line 547)May include surrounding context.

md
</div>
      </div>
      
      <!-- 方案3:更省时/更稳 -->
      <div class="plan-card plan-3">
        <div class="plan-header">
          <h3>🥉 方案3</h3>

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · reference/html-template.md (reported line 547)May include surrounding context.

md
</div>
      </div>
      
      <!-- 方案3:更省时/更稳 -->
      <div class="plan-card plan-3">
        <div class="plan-header">
          <h3>🥉 方案3</h3>

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

L003 的触发词列表包含“帮我搜机票”“机票怎么买最便宜”等宽泛自然语言,和普通旅行咨询高度重叠,缺少明确约束说明何时应进入“多目的地跨城优化”而非一般机票查询。这会增加在非目标场景下误调用该技能的概率。

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill directs the system to read and update persistent memory with travel preferences, departure city, and companion information, but it does not clearly obtain informed user consent for storing that personal data. Persistent profiling increases privacy risk because seemingly routine travel details can reveal home location, habits, and associations over time.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
99% confidence
Finding

The skill explicitly instructs disabling TLS certificate verification via NODE_TLS_REJECT_UNAUTHORIZED=0 before making flight-search requests. This removes protection against man-in-the-middle attacks and spoofed endpoints, allowing tampering with returned pricing or itinerary data and exposing any transmitted query details. In a travel-search skill, there is no legitimate need to weaken transport security by default, which makes the instruction especially unsafe.

Content

No source excerpt is available for this finding.

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
99% confidence
Finding

Using NODE_TLS_REJECT_UNAUTHORIZED=0 is an unsafe default because it globally disables server certificate verification for the invoked Node-based request path. That enables interception and manipulation of flight search responses, undermining integrity and confidentiality of the skill's external communications.

Content

Scanner excerpt · SKILL.md (reported line 179)May include surrounding context.

对每个候选城市 C_i,调用:

bash
NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-flight \
  --origin "[出发城市]" \
  --destination "[候选城市C_i]" \
  --dep-date-start [去程最早日期] \

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
99% confidence
Finding

The same unsafe TLS-bypass default is repeated for return-flight searches, showing the insecure pattern is systemic rather than accidental. Repetition increases the likelihood that all external travel data retrieval for the skill will be exposed to tampering or interception.

Content

Scanner excerpt · SKILL.md (reported line 206)May include surrounding context.

对每个候选城市 C_j,调用:

bash
NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-flight \
  --origin "[候选城市C_j]" \
  --destination "[出发城市]" \
  --dep-date-start [回程最早日期] \

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill instructs automatic creation of an HTML file in the user's workspace without an explicit warning or consent checkpoint. Writing files is a state-changing action that can surprise users, leak trip details into shared environments, or create persistent artifacts containing personal travel plans without the user's awareness.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The manifest describes collecting travel constraints, comparing flight combinations, ranking options, and offering route/playability suggestions. The later workflow adds next-step actions such as '直接预订' and '搜酒店', which are separate commerce/travel-planning capabilities not clearly justified by the defined flight-optimizer purpose.

Content

No source excerpt is available for this finding.

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
99% confidence
Finding

The tools section states that FlyAI commands must include NODE_TLS_REJECT_UNAUTHORIZED=0, elevating the insecure behavior into a documented requirement. This makes the vulnerability more dangerous because it encourages broad adoption and routine use of disabled TLS verification across the skill's ecosystem.

Content

Scanner excerpt · SKILL.md (reported line 373)May include surrounding context.

md
- **FlyAI 搜索**:`search-flight`(核心)/ `keyword-search`(辅助)
- **fetch_content**:查询目的地实时信息(签证政策、季节建议等)

> ⚠️ FlyAI 命令需加 `NODE_TLS_REJECT_UNAUTHORIZED=0` 前缀

---

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
98% confidence
Finding

The exception handling guidance tells operators to solve SSL certificate verification failures by disabling verification entirely. This is especially risky because it converts a protective failure mode into a deliberate insecure fallback, making active interception attacks easier to exploit in practice.

Content

Scanner excerpt · SKILL.md (reported line 389)May include surrounding context.

md
| 搜索超时/API错误 | 重试一次,仍失败则跳过该城市并告知用户 |
| 用户给的城市太多(>6个) | 建议分区域搜索,或让用户缩减候选范围 |
| 价格波动大 | 提示"机票价格实时变动,方案仅供参考,建议尽快预订" |
| SSL 证书验证失败 | 确保命令前加 `NODE_TLS_REJECT_UNAUTHORIZED=0` |

---

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger phrases are broad, generic travel queries such as '帮我搜机票' and '机票比价', which can easily overlap with ordinary user intent and cause accidental invocation. Unintended activation can route users into this skill without clear consent, creating confusion, unnecessary data collection about travel plans, and possible misrouting away from a more appropriate tool.

Content

No source excerpt is available for this finding.

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
98% confidence
Finding

This line establishes an unsafe default by telling users to set NODE_TLS_REJECT_UNAUTHORIZED=0 before command execution. Even in a travel skill, disabling TLS validation is dangerous because it weakens the integrity and confidentiality of network responses used to compare prices and present booking links.

Content

Scanner excerpt · reference/flyai-commands.md (reported line 3)May include surrounding context.

md
# FlyAI 命令参考

> ⚠️ **重要**:所有命令执行前需加 `NODE_TLS_REJECT_UNAUTHORIZED=0` 解决 SSL 证书验证问题

## 目录

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
97% confidence
Finding

The primary command format embeds TLS verification disabling directly into the canonical usage example. This normalizes insecure execution and increases the chance that every downstream use of the skill will occur without certificate validation.

Content

Scanner excerpt · reference/flyai-commands.md (reported line 24)May include surrounding context.

命令格式:

bash
NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-flight \
  --origin "[出发城市]" \
  --destination "[目的地城市]" \
  --dep-date [出发日期] \

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
96% confidence
Finding

This example repeats the insecure TLS-bypass setting for outbound flight search. Repetition across examples reinforces the unsafe behavior as standard practice and broadens exposure to manipulated or intercepted search results.

Content

Scanner excerpt · reference/flyai-commands.md (reported line 74)May include surrounding context.

bash
# 搜索去巴塞罗那的最便宜航班
NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-flight \
  --origin "杭州" --destination "巴塞罗那" \
  --dep-date-start 2026-10-01 --dep-date-end 2026-10-02 \
  --sort-type 3

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
96% confidence
Finding

This search example includes a global TLS verification bypass, making insecure transport appear required for ordinary price comparison queries. An attacker on the network path could alter returned pricing, flight details, or links shown to the user.

Content

Scanner excerpt · reference/flyai-commands.md (reported line 80)May include surrounding context.

md
--sort-type 3

# 搜索去马德里的最便宜航班
NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-flight \
  --origin "杭州" --destination "马德里" \
  --dep-date-start 2026-10-01 --dep-date-end 2026-10-02 \
  --sort-type 3

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
96% confidence
Finding

The example for searching flights to Lisbon also disables TLS verification. Although it is only documentation, it directly instructs unsafe execution that can undermine trust in network-fetched itinerary and booking data.

Content

Scanner excerpt · reference/flyai-commands.md (reported line 86)May include surrounding context.

md
--sort-type 3

# 搜索去里斯本的最便宜航班
NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-flight \
  --origin "杭州" --destination "里斯本" \
  --dep-date-start 2026-10-01 --dep-date-end 2026-10-02 \
  --sort-type 3

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
96% confidence
Finding

This return-flight example again propagates NODE_TLS_REJECT_UNAUTHORIZED=0 as normal usage. Because booking and price-comparison flows rely on external HTTPS responses, disabling validation creates an unnecessary channel for response spoofing or tampering.

Content

Scanner excerpt · reference/flyai-commands.md (reported line 96)May include surrounding context.

bash
# 搜索从巴塞罗那回杭州的最便宜航班
NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-flight \
  --origin "巴塞罗那" --destination "杭州" \
  --dep-date-start 2026-10-06 --dep-date-end 2026-10-07 \
  --sort-type 3

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
96% confidence
Finding

This example instructs users to search Madrid-to-Hangzhou flights with TLS verification disabled. The skill context makes this especially inappropriate because users may rely on the authenticity of prices, schedules, and booking URLs when making purchase decisions.

Content

Scanner excerpt · reference/flyai-commands.md (reported line 102)May include surrounding context.

md
--sort-type 3

# 搜索从马德里回杭州
NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-flight \
  --origin "马德里" --destination "杭州" \
  --dep-date-start 2026-10-06 --dep-date-end 2026-10-07 \
  --sort-type 3

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
96% confidence
Finding

This itinerary example perpetuates the unsafe default of disabling TLS certificate checks. Repeated insecure examples increase the likelihood that users or downstream agents will copy-paste the pattern into automation and operational workflows.

Content

Scanner excerpt · reference/flyai-commands.md (reported line 108)May include surrounding context.

md
--sort-type 3

# 搜索从里斯本回杭州
NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-flight \
  --origin "里斯本" --destination "杭州" \
  --dep-date-start 2026-10-06 --dep-date-end 2026-10-07 \
  --sort-type 3

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
96% confidence
Finding

The direct-flight search example still disables TLS verification, showing the issue is systemic rather than incidental. Embedding insecure transport into filtered search scenarios provides no functional need tied to the business purpose of the skill.

Content

Scanner excerpt · reference/flyai-commands.md (reported line 118)May include surrounding context.

bash
# 仅直飞
NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-flight \
  --origin "杭州" --destination "巴塞罗那" \
  --dep-date 2026-10-01 \
  --journey-type 1 --sort-type 3