Security audit
Security checks for vulnerabilities and agentic risk
The main skill is an international-relations analysis prompt, but the package also ships unrelated executable file-processing scripts that are not disclosed by the skill description.
Review this before installing because the IR analysis prompt is accompanied by unrelated local-file utilities. They do not appear malicious or automatically invoked, but the publisher should either remove them or clearly document why they are included and when they should be run.
声明描述的是一个面向国际关系与比较政治内容分析的技能，即应当提供外交/安全/身份政治等领域的实质性分析能力。但实际代码并不进行任何国际关系推理、比较政治分析或地缘战略评估；它只是一个研究资料合并与摘要生成工具。其主要行为是读取本地 markdown 文件、统计 URL 和文本标记、提取标题、检测显式“矛盾”词，并输出汇总表。这与声明的核心用途在主目的和能力上都明显不一致，因此应判定为描述与行为不匹配。
There is a clear description-behavior mismatch. The declared purpose says the skill supports international relations and comparative politics analysis. However, the supplied code does not implement any IR reasoning, policy analysis, identity analysis, or geopolitical evaluation. Its primary and only observable function is validating generated SKILL.md files against Phase 4 quality criteria. This is a materially different purpose, not a supporting implementation detail of an IR analysis skill.
There is a clear description-behavior mismatch. The declared purpose describes a substantive international relations analysis skill for geopolitical and comparative politics tasks. However, the supplied code performs only subtitle cleaning and transcript generation from SRT/VTT files. Its primary purpose, inputs, outputs, and operational domain are entirely different from IR analysis. No part of the code implements international relations reasoning, policy analysis, comparative political analysis, or related domain-specific functionality.
The file implements subtitle conversion and local file I/O even though the declared skill is for international-relations analysis. This capability mismatch is dangerous because unrelated file-manipulation utilities increase the agent's attack surface and can be used to justify unexpected access to local user files under a misleading skill identity.
Without declared permissions the skill's intent is opaque and cannot be validated.
The manifest describes a skill for international relations analysis, but this file is a utility that merges six agent research markdown files into a summary table for a review checkpoint. Its code performs local file aggregation and heuristic markdown parsing for a research workflow, which does not implement or directly support the declared IR-analysis capability in any domain-specific way.
The script claims to automatically verify whether a generated SKILL.md passes 'Phase 4 quality standards', but the concrete checks are for items such as '心智模型', '表达DNA', '诚实边界', and '一手来源占比'. For a skill described as an international-relations analysis assistant, these are generic persona/template QA constraints rather than behavior or capability checks tied to international-relations analysis, creating a semantic mismatch between the stated skill purpose and what this code actually validates.
Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.
python3 quality_check.py <SKILL.md路径>
示例:
python3 quality_check.py .claude/skills/elon-musk-perspective/SKILL.md
"""
import sys
Within an international-relations analysis context, adding subtitle format conversion is functionally unnecessary and broadens available capabilities beyond the user's likely expectations. Even if the script is not overtly malicious, irrelevant file-conversion code can facilitate data handling outside the declared mission and makes misuse or covert staging of local content easier.
Data from a source is assigned to a variable that is later passed to a sink, creating a variable-mediated taint flow.
else:
transcript = clean_srt(content)
output_path.write_text(transcript, encoding='utf-8')
# 统计
word_count = len(transcript)
This file contains natural-language instructions, usage text, and output messages only in Chinese, including the module docstring and runtime prompts. Under the policy rule, forcing a specific language without opt-in or a documented region-specific reason is a natural-language policy violation.
This code file contains user-facing natural-language strings and documentation entirely in Chinese, including the module docstring and usage example. Under the language/locale policy rule, forcing a specific language without user opt-in can be a policy violation when no alternative or justification is provided.
The script prints usage, error, and result messages only in Chinese during execution. Because the file does not offer a locale choice or explain a justified region-specific constraint, this appears to enforce a single language for end users.
No suspicious patterns detected.