Literature Search Skill

学术文献搜索技能 - 支持多平台文献搜索、内容提取和分析

Security Scan

The skill is an instruction-only literature search helper and its declared requirements and runtime instructions are largely coherent; only a minor mismatch (declared python/curl requirement not used in the SKILL.md) was found.

Purpose & Capability

The skill's name/description (multi-platform academic search and analysis) matches the instructions (use web_search/web_fetch, then summarize/analyze). However the metadata declares a requirement of either 'python' or 'curl' (anyBins) even though the SKILL.md exclusively references agent actions web_search/web_fetch and does not show any use of python or curl. This is a minor mismatch (likely harmless) but unnecessary binaries are requested.

Instruction Scope

SKILL.md stays on-topic: it instructs the agent to search, fetch, summarize, classify, and generate reports for academic literature. It does not ask the agent to read arbitrary local files, environment variables, or unrelated system state. Note: the instructions are high-level (e.g., '模型自动进行内容分析和分类') and rely on the agent's web_fetch to retrieve URLs; web_fetch can fetch arbitrary web content, so the agent's autonomy could be misused to fetch non-literature content if prompted to do so.

Install Mechanism

There is no install spec and no code files — this is instruction-only, so nothing is written to disk and no external packages are pulled. This is the lowest-risk install model.

Credentials

The skill requests no environment variables or credentials, which is appropriate for a search-and-summarize tool. (Some target platforms like IEEE/ACM/Google Scholar may require user accounts or institution access in practice, but the skill does not request those credentials — which is proportionate.)

Persistence & Privilege

always is false and the skill does not request elevated or persistent presence. disable-model-invocation is false (normal), meaning the agent may call the skill autonomously — this is expected and not itself a concern given the rest of the footprint.

Assessment

This skill appears coherent and instruction-only, but review these points before installing: 1) The metadata requires either python or curl although the instructions use agent actions (web_search/web_fetch) — confirm whether those binaries are actually needed. 2) web_fetch can retrieve any URL; avoid giving the agent sensitive links or credentials and monitor network activity if possible. 3) The skill does not request credentials for paywalled platforms; if you plan to use those, prefer providing credentials only via secure, explicit means and verify the agent won't exfiltrate them. 4) Because the SKILL.md is high-level, test the skill with non-sensitive queries first to ensure it behaves as expected.

学术文献搜索技能

专业的学术文献搜索和分析工具，支持多个学术平台。

支持的学术平台

• arXiv - 预印本论文
• Google Scholar - 学术搜索引擎
• PubMed - 生物医学文献
• IEEE Xplore - 工程和计算机科学
• ACM Digital Library - 计算机科学

使用方法

基础搜索

# 搜索特定主题
web_search query:"[your topic] site:arxiv.org" count:10 freshness:"month"

# 搜索特定期刊
web_search query:"[your topic] site:nature.com" count:5

高级搜索技巧

时间范围搜索

# 最近1个月
web_search query:"large language model optimization" freshness:"month"

# 最近1年
web_search query:"quantum computing applications" freshness:"year"

多关键词搜索

# 组合搜索
web_search query:"(machine learning) AND (healthcare) site:arxiv.org" count:10

文献分析流程

1. 文献搜索

搜索相关主题的文献，获取标题、摘要和链接。

2. 内容获取

使用 web_fetch 获取文献全文内容。

3. 内容分析

利用模型能力进行：

• 内容总结 - 提取核心观点
• 创新点识别 - 识别技术突破
• 分类整理 - 按研究方向分类
• 差异对比 - 比较不同方法

4. 报告生成

生成结构化的文献综述报告。

示例工作流

完整的文献调研

# 1. 搜索文献
web_search query:"transformer optimization techniques" count:8 freshness:"month"

# 2. 获取重点文献全文
web_fetch url:"https://arxiv.org/abs/important-paper"

# 3. 分析总结
# 模型自动进行内容分析和分类

输出格式

文献摘要

• 标题和作者
• 发表时间
• 核心贡献
• 技术方法
• 实验结果

分类报告

• 按研究方向分类
• 创新点对比
• 发展趋势分析
• 研究空白识别

注意事项

1. 版权尊重 - 仅用于学术研究
2. 引用规范 - 正确引用来源
3. 数据备份 - 保存搜索结果
4. 网络限制 - 部分平台可能有访问限制