Security audit
Security checks for vulnerabilities and agentic risk
This is a Chinese-language reference and guidance skill for comparing AI coding tools, with no evidence of hidden execution, persistence, data theft, or destructive behavior.
Install this as an informational Chinese-language reference skill. Treat its examples about shell commands, deployment, file modification, indexing, and persistent context as descriptions of other AI coding tools, and keep normal user review in place before allowing any real coding agent to make changes or run commands.
声明描述的是一个面向用户的“AI 编程工具分析技能包”，核心用途应是分析、比较、学习工具以及提供 Agent/提示词相关指导。然而代码并未实现任何此类分析逻辑，也没有处理用户查询、工具评估、对比、推荐或提示词生成。相反，这段代码的主要功能是读取本地技能元数据和文档，统计内容，写出 package_info JSON，并创建 ZIP 压缩包。该行为属于构建/打包辅助脚本，和声明的主要用途存在实质性偏差，因此应判定为描述与实际行为不匹配。
The declared description presents this as an end-user skill suite for AI coding tool analysis, comparison, agent system design, and prompt engineering. However, the supplied code chunk is only a validation utility for the skill package itself. It inspects local files under a hardcoded directory, parses SKILL.md and _meta.json, checks naming conventions and required fields, and reports pass/fail results. This is a materially different primary purpose from the declared functionality. While such a validator could be a development support script within the package, the actual code shown does not implement the described analytical or educational capabilities, so the description does not accurately represent what this code chunk actually does.
Without declared permissions the skill's intent is opaque and cannot be validated.
The skill description defines applicability in very broad terms without clear boundaries, which makes over-triggering and misrouting more likely. In agent systems, vague applicability expands the prompt surface area and can cause the skill to engage on unrelated inputs, potentially exposing context or producing unauthorized specialized behavior.
文件整体描述与交互说明均固定为中文，且未说明可根据用户偏好切换语言，也没有提供语言/地区选择入口。若组织要求不得在未经用户同意的情况下强制特定语言，这种默认单一语言设定构成自然语言策略风险。
The skill supports very broad natural-language activation, including generic phrases like directly describing a need, which can cause accidental invocation during ordinary conversation. Overbroad triggering can unexpectedly route user prompts into a specialized workflow, increasing the chance of unintended data processing, confusion, or prompt-scope hijacking by unrelated text.
The routing table uses generic keywords such as 'compare', 'learn', 'guide', and 'prompt'-style concepts without sufficient contextual constraints. This broad matching can be exploited or accidentally triggered by benign user text, causing unintended skill routing and making prompt injection or context confusion easier within a multi-skill environment.
The manifest description is written entirely in Chinese, which indicates a language-specific presentation without any accompanying language choice, opt-in, or justification that the skill is region-specific. This can violate organizational language/locale policy when users are not given an explicit option to use another language.
This markdown file documents capabilities including create_file, delete_file, and run_in_terminal, and later states the agent can automatically explore and modify code. Although it mentions safety principles, it does not clearly warn users that the skill may delete files or execute shell commands that affect the local system.
This markdown file documents the skill as capable of creating or overwriting files via Write and executing shell commands via Bash, both of which can affect user data or system state. Although the document mentions transparency and an Ask mode, it does not provide an explicit warning to users about the risks or potential impact of these operations in the skill description.
The file states that the system automatically collects current file contents, cursor position, edit history, diagnostics, and Git status. These behaviors may affect user privacy or expose sensitive project context, but the description does not warn users that such data will be accessed automatically.
The manifest frames the skill as an analysis and learning suite for AI coding tools plus agent/prompt design support. Documenting a shell tool for executing commands and managing processes describes agent execution functionality, which is not an obvious requirement for a skill whose declared purpose is comparative analysis and design reference.
This markdown file documents safety-relevant operations including shell command execution, process control, file creation/editing, deployment, and exposing local ports, but it does not include any cautionary warning about effects on user systems, data, or network exposure. Under the markdown-specific SQP-2 criteria, descriptions of behaviors that could affect system integrity or privacy should disclose those risks to users.
The manifest describes this skill package as being for analyzing, comparing, learning AI coding tools, designing agent systems, and prompt engineering. This file documents operational capabilities such as deploying frontend/backend services and exposing local ports, which are execution capabilities rather than analysis/reference behavior and are not justified by the stated purpose of a reference/analysis skill.
This markdown file documents tools that can write or edit files (Write, Edit) and execute shell commands (Bash), but it does not include any warning or cautionary note about modification of user files or command-side effects. Under the markdown-specific SQP-2 criteria, descriptions of behaviors that can affect user data or system integrity should disclose those risks.
The document states that system context may include operating system information, current directory, shell environment, and user information, and that project context is persisted across sessions. This is a data- and privacy-relevant behavior, but the description does not warn users that such context may be collected and retained.
This markdown guidance forces a specific language/locale for all readers through its title and full-body content, and nowhere indicates that Chinese is optional or that the document is intended only for a Chinese-speaking audience. Under the stated policy, a fixed language without user opt-in or documented justification is a natural-language policy violation.
The file specifies exact English approval phrases for progressing through workflow gates even though the surrounding document is in Chinese. This imposes a specific language on the user and may exclude users who operate in another locale.
This is a markdown file, so SQP-2 applies to descriptions that omit warnings about behaviors affecting system integrity. The section describes a tool for executing Shell commands and lists parameters and outputs, but it provides no user-facing caution about command safety, side effects, or the need for confirmation before potentially destructive execution.
SQP-3 applies to all file types and covers natural-language policy issues such as forcing a specific language without user opt-in. This markdown reference is fully written in Chinese and does not indicate that the language is optional, user-selected, or justified by a region-specific context.
A skill document that effectively forces one language can violate language/locale policy when it does not offer an explicit choice or state that the locale is intentionally limited. Here, the content is presented only in Chinese with no indication that users may choose another language.
This reference file is primarily written in Chinese, but it prescribes English-only interaction patterns such as the user-story format and later approval prompts. That can violate a language/locale policy when users are not given an explicit option to use their preferred language.
SQP-3 applies to all file types and includes language or locale policy violations. This file presents all operational descriptions in Chinese and does not offer an opt-in, alternative language, or justification for a Chinese-only constraint, which can amount to forcing a specific language without user choice.
This markdown file presents all instructions and examples exclusively in Chinese, starting from the title and continuing throughout the document. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale restriction is explicitly justified, which is not present here.
This code includes user-facing natural-language content in Chinese in the module docstring, comments, and console output context, but does not indicate that the skill is intentionally Chinese-only or provide any user opt-in for locale. That can violate a language/locale policy when users may expect language-neutral or selectable messaging.
No suspicious patterns detected.
