Back to skill

Security audit

同款打卡

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent travel-planning helper, but it asks agents to install mutable global software, sometimes with sudo, disables TLS verification for booking-related searches, and reads or stores travel profiles with broad persistence.

Review carefully before installing. Do not let the skill run global npm installs or sudo automatically, do not use NODE_TLS_REJECT_UNAUTHORIZED=0, and verify any flight, hotel, or ticket booking link in a trusted browser session before paying. Only allow profile memory or ~/.flyai/user-profile.md storage if you want cross-session personalization and are comfortable storing travel preferences and family details.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (3)

T08 · Insecure Dependencies

Error
Location
SKILL.md:41
Finding

Unpinned Global Dependency Installation with Optional Root Privileges

Content
View full analysis
Remediation
View remediation
``` 9. Review the package contents and lifecycle scripts before approving a new version. 10. Execute the CLI in a sandbox with minimum filesystem, process, and network privileges. ]]>

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:107
Finding

TLS Certificate Verification Disabled for Travel Searches

Content
View full analysis
``` The unsafe configuration is applied to multiple operations: ```bash NODE_TLS_REJECT_UNAUTHORIZED=0 flyai keyword-search --query "[work name] filming location [primary filming city]" ``` ```bash NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-poi --city-name "[filming city]" --keyword "[specific location name]" ``` ```bash NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-flight --origin "[departure city]" --destination "[nearest international airport]" --dep-date [departure date] --back-date [return date] --sort-type 3 ``` ```bash NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-flight --origin "[departure city]" --destination "[destination]" --dep-date [departure date] --back-date [return date] --sort-type 3 ``` ```bash NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-hotel --dest-name "[destination]" --check-in-date [check-in date] --check-out-date [check-out date] ``` ### Technical Analysis Setting `NODE_TLS_REJECT_UNAUTHORIZED=0` disables TLS certificate validation for the entire Node.js process. The client will therefore accept expired, self-signed, hostname-mismatched, or attacker-generated certificates. Encryption without server authentication does not protect against an active man-in-the-middle attacker. Because the skill uses the affected commands to retrieve flight, hotel, attraction, pricing, and booking-link data, a network attacker may forge service responses that appear legitimate to the agent. The skill subsequently requires `jumpUrl` and `detailUrl` values from these responses to be displayed as booking links. A forged response could therefore introduce attacker-controlled destinations into the final user-facing itinerary. ### Attack Path 1. The agent starts FlyAI with `NODE_TLS_REJECT_UNAUTHORIZ ...[truncated 1407 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:116
Finding

Shell Command Injection Through User-Controlled Travel Parameters

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Rogue AgentSelf-Modification, Session Persistence
Findings (31)

Context-Inappropriate Capability

Critical
Category
Not specified by scanner
Confidence
100% confidence
Finding

The skill explicitly instructs disabling TLS certificate verification with NODE_TLS_REJECT_UNAUTHORIZED=0 for command execution. That defeats HTTPS trust checks and enables man-in-the-middle interception or tampering of flight, hotel, POI, or package data returned to the agent, which is especially dangerous in a booking-oriented workflow.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill requires installing and globally upgrading an external CLI before performing its travel-planning task, and even suggests elevated installation paths. This expands the skill from itinerary generation into modifying the host environment and executing network-fetched code, which creates unnecessary supply-chain and system-integrity risk for a content/planning workflow.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
100% confidence
Finding

The skill normalizes bypassing SSL certificate verification and provides no user-facing warning about the security consequences. In a skill that surfaces booking links and travel results, this can lead to silent acceptance of tampered network responses and unsafe downstream decisions.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill directs startup access to stored user memory and a local profile file even before establishing per-request necessity. For a travel recommendation skill, broad profile retrieval increases privacy exposure and data collection beyond the minimum needed to answer a single itinerary request.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill reads a local user profile file at startup without clearly disclosing this access in the skill description or obtaining contextual consent. This is a privacy and transparency issue because the file may contain unrelated or sensitive personal travel preferences.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

Automatically loading user history and local profile data at startup lacks data minimization and per-request justification. In context, the skill can plan an itinerary with direct user prompts, so unconditional profile access unnecessarily broadens exposure of personal data.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill instructs global package installation and suggests sudo without clearly warning that this modifies the system and may run with elevated privileges. For a travel-planning skill, such host changes are unnecessary and increase the chance of unintended system impact or abuse via compromised package sources.

Content

No source excerpt is available for this finding.

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
96% confidence
Finding

The skill recommends sudo for npm global installation, which elevates execution privileges for externally sourced package installation. This increases the blast radius of any package compromise or installation script abuse and is not justified by the business purpose of a pilgrimage itinerary assistant.

Content

Scanner excerpt · SKILL.md (reported line 56)May include surrounding context.

md
| 情况 | 处理方式 |
|-----|---------|
| npm 未安装 | 提示用户先安装 Node.js (https://nodejs.org/) |
| 权限不足 | 建议使用 `sudo npm install -g @fly-ai/flyai-cli@latest --registry=https://registry.npmjs.org` 或使用 nvm 管理 Node |
| 网络问题 | 建议用户检查网络或使用国内镜像 `npm config set registry https://registry.npmmirror.com` |

**注意:** 此步骤只在首次使用时执行,后续调用会直接跳过已安装的情况。

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
100% confidence
Finding

Setting NODE_TLS_REJECT_UNAUTHORIZED=0 is an unsafe default because it disables certificate validation for TLS connections. This allows interception or tampering of responses from remote services and undermines the trustworthiness of any travel data or links presented to the user.

Content

Scanner excerpt · SKILL.md (reported line 111)May include surrounding context.

重要:SSL 证书验证问题处理 如果遇到 "SSL 证书验证失败" 错误,需要在命令前加上环境变量:

bash
NODE_TLS_REJECT_UNAUTHORIZED=0 flyai <command>

2.1 极速搜索取景地信息

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
100% confidence
Finding

The keyword-search example embeds a TLS-verification bypass directly into the primary search workflow. Because this is presented as the normal command pattern, it encourages insecure operation by default and exposes all retrieved data to manipulation.

Content

Scanner excerpt · SKILL.md (reported line 116)May include surrounding context.

2.1 极速搜索取景地信息

bash
NODE_TLS_REJECT_UNAUTHORIZED=0 flyai keyword-search --query "[作品名称] 取景地 [主要取景城市]"

2.2 结合 AI 知识补充

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
100% confidence
Finding

The POI search command also bakes in TLS verification disablement, making insecure network use part of routine location validation. That is particularly risky because the skill relies on these results to validate real-world destinations and booking context.

Content

Scanner excerpt · SKILL.md (reported line 132)May include surrounding context.

对于每个取景地城市,调用 FlyAI 搜索具体 POI:

bash
NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-poi --city-name "[取景地城市]" --keyword "[具体地点名]"

POI 验证要点:

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
100% confidence
Finding

The international flight search command uses disabled TLS validation while retrieving travel offers and links. Compromised responses in this context could mislead users about destinations, prices, or booking URLs and have direct financial consequences.

Content

Scanner excerpt · SKILL.md (reported line 144)May include surrounding context.

国际航线:

bash
NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-flight --origin "[出发城市]" --destination "[最近国际机场]" --dep-date [出发日期] --back-date [返程日期] --sort-type 3

国内航线:

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
100% confidence
Finding

The domestic flight search example repeats the insecure transport default, reinforcing a dangerous pattern across booking operations. Because the skill later surfaces booking links, tampered results could redirect users or corrupt pricing and itinerary information.

Content

Scanner excerpt · SKILL.md (reported line 149)May include surrounding context.

国内航线:

bash
NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-flight --origin "[出发城市]" --destination "[目的地]" --dep-date [出发日期] --back-date [返程日期] --sort-type 3

第5步:酒店搜索

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
100% confidence
Finding

The hotel search workflow similarly disables TLS verification, extending insecure transport to accommodation lookup. In a travel skill that extracts detail URLs for user action, this raises the risk of link manipulation and deceptive booking output.

Content

Scanner excerpt · SKILL.md (reported line 157)May include surrounding context.

优先搜索靠近主要打卡点的酒店:

bash
NODE_TLS_REJECT_UNAUTHORIZED=0 flyai search-hotel --dest-name "[目的地]" --check-in-date [入住日期] --check-out-date [退房日期]

查看更多酒店参数选项:

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
99% confidence
Finding

The failure-handling table explicitly recommends the TLS-bypass environment variable as the fix for certificate errors. That codifies insecure behavior as an operational fallback, making the overall skill more dangerous because users or agents are guided to prefer availability over authenticity.

Content

Scanner excerpt · SKILL.md (reported line 277)May include surrounding context.

md
| 国际目的地签证复杂 | 提醒签证要求,推荐调用 flyai 签证相关搜索 |
| 某些取景地需要预约/许可 | 标注"⚠️ 需提前预约",提供预约渠道 |
| FlyAI 返回空结果 | 尝试调整搜索词或使用 AI 知识补充 |
| SSL 证书验证失败 | 使用 `NODE_TLS_REJECT_UNAUTHORIZED=0` 环境变量 |

## 扩展能力

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The skill proposes learning from user-provided new IP information and recording it into a knowledge base, which extends behavior beyond one-time planning. This creates data-retention and prompt/data-poisoning risks if untrusted user content is stored without validation or consent.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

Instructing the skill to learn from user-supplied information and record it into a knowledge base introduces storage of untrusted content. That can create privacy issues and contamination of future outputs if the stored material is inaccurate, malicious, or copyrighted and is later treated as trusted knowledge.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The manifest says this skill is for ‘影视/综艺/动漫同款打卡·圣地巡礼’, starting from a work and matching filming-location POIs into a pilgrimage route. However, this file documents ai-search as a generic semantic travel search supporting hotels, attractions, flights, trains, budgets, companions, and mixed travel intent, with examples for ordinary Hangzhou and Tokyo trip planning unrelated to filming-location pilgrimage use cases.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The example dialogue is entirely written in Chinese and presents the interaction as the default behavior, with no indication that users may choose another language. This can violate language/locale policy when a skill implicitly enforces a specific language without user opt-in.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The document defines cross-session persistence of a broad 'travel profile' for a skill whose stated purpose is generating pilgrimage/check-in routes from a referenced film/show/anime. Persisting user profile data beyond what is necessary creates unnecessary data retention and expands the privacy attack surface if the data is later accessed, leaked, or reused outside the immediate request.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The Qoder Memory instructions direct the agent to store personal data in an external memory system without clearly warning the user that their profile will be transmitted to and retained by that system. This creates a transparency and privacy problem because users may not expect cross-session storage in a separate platform component.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The file-mode instructions tell the agent to save profile data locally but provide no privacy warning, consent language, or notice that the information will persist across sessions on disk. Local plaintext storage of personal preferences and family/travel details can be read by other local users, software, backups, or logs, making the persistence materially risky.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
84% confidence
Finding

Creating ~/.flyai/ establishes session persistence on the local filesystem for user profile data, which is a real privacy and data-retention concern in this context even though the command itself is not malicious. In a narrowly scoped pilgrimage-route skill, automatic creation of a persistent local store makes the skill more dangerous because it normalizes silent retention of personal information beyond the immediate interaction.

Content

Scanner excerpt · reference/user-profile-storage.md (reported line 73)May include surrounding context.

如果 ~/.flyai/ 目录不存在,需要先创建:

bash
mkdir -p ~/.flyai

文件格式

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The prescribed fields include family composition, child age, travel history, and other personal attributes that are not clearly required to answer 'same-location check-in' or filming-location itinerary questions. Collecting and storing unnecessary personal data increases privacy risk and can expose sensitive household details if the memory store or local file is accessed by other tools, users, or processes.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
85% confidence
Finding

The only usage examples are Chinese-language queries, which can imply the skill is intended to be used in a specific language without any stated opt-in, language choice, or regional limitation. Because this is a natural-language reference file and no justification for the language constraint is documented, it presents a minor locale-policy concern.

Content

No source excerpt is available for this finding.