Patent Scout

通过百度或Google Patents在线检索中国专利信息，支持关键词和专利号查询，输出结构化专利摘要和申请人等数据。

Security Scan

The skill's code, instructions, and dependencies match its stated purpose (web-scraping Chinese patents via Baidu or Google Patents); nothing indicates hidden credential access or unrelated capabilities, though there are minor inconsistencies in versions and paths you should be aware of.

Purpose & Capability

Name/description describe online patent search; the bundle includes a Node.js script (axios + cheerio) and CLI instructions to query Baidu/Google Patents which is exactly what's needed for that purpose. Dependencies and required binaries (Node/npm at runtime) are proportionate. Minor inconsistencies: package.json version is 1.0.0 while scripts and _meta.json reference 2.0.0 and registry metadata shows 1.0.1 — and SKILL.md's install path (cd internal/java/patent-scout) doesn't match the file layout. These look like packaging/editorial issues, not functional mismatches.

Instruction Scope

SKILL.md instructs running the included script with flags, using --proxy for Google Patents, and exporting results to a file — all within the patent-search scope. The instructions do not request or read unrelated system files, environment secrets, or non-user-provided credentials. The script performs network requests to external patent sites (expected for this skill).

Install Mechanism

No formal install spec in the registry (instruction-only), but package.json and package-lock.json are included and SKILL.md recommends npm install. Dependencies are standard (axios, cheerio, commander, https-proxy-agent) and package-lock entries reference the npmmirror registry (a common China npm mirror). Because there is no curated install spec, installing requires running npm install locally — which will fetch packages from the registry specified in package-lock.json.

Credentials

Skill declares no required environment variables or credentials. The code accepts a user-provided --proxy option (user-supplied) and sets a synthetic BAIDUID cookie header for requests — neither require secrets. There are no requests for unrelated tokens/keys and the script does not appear to read hidden config paths.

Persistence & Privilege

always is false and the skill does not request persistent or elevated system privileges. It writes output only to a user-specified file (if used). Autonomous model invocation is allowed (platform default) but not combined with any unusual privileges here.

Assessment

This skill appears to do what it claims: an npm/Node.js scraper for patent results from Baidu or Google Patents. Before installing or running it: (1) inspect the included scripts yourself (or run in an isolated environment) because npm install will fetch third‑party packages; (2) note small metadata inconsistencies (version numbers and an odd install path in SKILL.md) — not malicious but worth confirming you have the correct release; (3) be careful when using a proxy: provide only proxies you trust (an attacker-controlled proxy could see all traffic you send through it); (4) respect target sites' robots/TOS and throttle requests (SKILL.md already warns about rate limits); (5) if you need higher assurance, run npm install with a lockfile audit or use a sandbox/container to execute the tool.

Patent Scout - 中国专利搜索 Skill

真实的中国专利搜索工具，通过百度搜索或 Google Patents 在线检索中国专利信息。

触发条件

当用户提到以下内容时使用此 Skill：

• "搜索专利"、"查专利"、"专利检索"
• "patent search"、"find patents"
• "技术专利"、"发明专利"、"实用新型"
• 提到具体的中国专利号（如 CN116789012A）
• 需要查看专利摘要、申请人等信息

功能特性

数据源

• 百度搜索（默认）— 国内直连，无需代理，适合中国网络环境
• Google Patents — 全球专利数据库，需代理访问

搜索能力

• 关键词搜索（中文）
• 专利号精确查询（CN 开头）
• 自动过滤广告和无关结果
• 自动提取专利号

输出格式

• Markdown（默认）— 结构化展示专利信息
• JSON — 方便程序处理

使用方法

关键词搜索

# 默认百度搜索
node scripts/patent-scout.js --query "工业防火墙"

# 限制结果数量
node scripts/patent-scout.js --query "深度学习" --limit 5

# 导出到文件
node scripts/patent-scout.js --query "固态电池" --output results.md

# JSON 格式输出
node scripts/patent-scout.js --query "5G" --format json

专利号查询

node scripts/patent-scout.js --patent-id CN116789012A

使用 Google Patents（需代理）

node scripts/patent-scout.js --query "芯片" -s google --proxy http://127.0.0.1:7890

参数说明

参数	说明	默认值
-q, --query	搜索关键词	-
-p, --patent-id	专利号（如 CN116789012A）	-
-l, --limit	结果数量	10
-s, --source	数据源 (baidu/google)	baidu
-o, --output	输出文件路径	-
-f, --format	输出格式 (markdown/json)	markdown
--proxy	代理地址（Google Patents 需要）	-

技术栈

• Node.js — 运行环境
• axios — HTTP 请求
• cheerio — HTML 解析
• commander — 命令行参数解析

安装

cd internal/java/patent-scout
npm install

示例场景

场景 1: 技术调研

用户: "帮我搜索关于工业防火墙的专利"
→ node scripts/patent-scout.js --query "工业防火墙" --limit 5

场景 2: 专利号查询

用户: "查一下专利 CN121567465A"
→ node scripts/patent-scout.js --patent-id CN121567465A

场景 3: 导出分析

用户: "搜索固态电池专利并导出"
→ node scripts/patent-scout.js --query "固态电池" --output results.md

注意事项

• 百度搜索为默认数据源，国内可直连
• Google Patents 在中国大陆需要配合代理使用
• 请控制请求频率，避免被限流
• 部分百度结果可能无法提取完整专利号

许可证

MIT License

作者

闫老师团队

更新日志

v2.0.0

• 重写为真实网络搜索（移除演示数据）
• 支持百度搜索（国内直连）
• 支持 Google Patents（需代理）
• 自动过滤广告
• 自动提取中国专利号

v1.0.0

• 初始版本（演示数据）