Wenshu Search - 裁判文书网搜索通用技能

通过裁判文书网账户登录，按案由、地区、年份、审判程序等条件搜索并提取判决书全文，支持Excel导出。

Security Scan

The skill appears to implement the claimed Wenshu search/scrape functionality, but the package metadata omits required runtime dependencies and the script asks for and uses your Wenshu login credentials — review the code and run only with a disposable account or in a sandbox.

Purpose & Capability

The name/description (search and extract judgments from 裁判文书网) match the script's behavior: UI-driven login, page navigation, text extraction, and Excel export. However the registry metadata lists no required binaries or dependencies while the script clearly expects Python packages (Crypto.Cipher, likely Playwright/Selenium, Excel libraries). That omission is inconsistent and important for security/operation.

Instruction Scope

SKILL.md and the script instruct interactive entry of your Wenshu account and password and then perform automated browsing and content extraction. The instructions claim pure UI simulation and forbid request interception; that matches the visible code comments. Still, the runtime will read/write a local .wenshu_config.json (stores username and filters; password is masked) and will send login credentials to the remote court site—which is required for the stated purpose but raises privacy/account-risk considerations.

Install Mechanism

There is no install spec (instruction-only) which reduces installer risk. But the included Python script imports third-party modules (e.g., Crypto.Cipher) and almost certainly relies on a browser automation package and browser binaries; these dependencies are not declared in the metadata or SKILL.md. Lack of declared dependencies is an oversight and makes safe installation harder (you must inspect and install packages yourself).

Credentials

The skill requests no environment variables in the registry, which is consistent, but it does require you to enter your Wenshu username and password at runtime. The script stores a local config file with username and search settings (password is not saved in plaintext). Requiring account credentials is proportionate to the stated purpose, but it is a sensitive action — the user should supply only an account they trust and be aware of potential account lock or terms-of-service issues.

Persistence & Privilege

always: false and no claims of global persistence. The script writes a local .wenshu_config.json in its directory (self-contained), which is normal for this use case. It does not request elevated system privileges or modify other skills' configurations.

What to consider before installing

This skill appears to do what it says (log in to 裁判文书网, browse, extract judgment text, and export to Excel), but exercise caution: 1) The bundle metadata does not declare required Python packages or browser automation dependencies — inspect the full script and install dependencies in a controlled environment. 2) The script asks for your Wenshu account password and will use it to log into the remote site; use a throwaway or non-critical account if you must run it. 3) Running large-scale scraping may violate the site's terms or trigger account blocks — follow the safety throttle noted in SKILL.md. 4) Review the entire wenshu_search.py for any network calls to external domains beyond the court site before running; if you are not comfortable auditing the code yourself, do not run it with your primary credentials. 5) Prefer running inside a sandbox/container and consider network monitoring to confirm no unexpected exfiltration occurs.

fanghongyuanwenshu — 裁判文书网搜索通用技能

版本: v10.1 | 更新: 2026-04-04
本技能为通用分享版，不含任何个人账号密码信息。 首次运行会引导补全账号密码和搜索条件，之后自动记忆。

---

功能概述

功能	说明
搜索方式	纯 UI 操作，不拦截不篡改任何请求
筛选条件	案由 / 文书类型 / 地区 / 年份 / 审判程序
正文提取	自动访问文书详情页，提取判决书全文
正文清洗	自动去除网页噪音（导航/版权/平台链接/板块标签）
Excel 导出	13列，含判决书全文，格式自动换行
安全机制	详情页间隔3秒，每5篇暂停5秒，连续2篇失败自动停
静默模式	默认 headless=True，后台运行不弹浏览器
账号记忆	首次输入后自动记忆，密码仅首次需要

---

搜索关键词建议

运行时会提示输入以下参数，以下为常用示例：

参数	说明	常用示例
案由	纠纷类型（中文全称）	房屋租赁合同纠纷、民间借贷纠纷、买卖合同纠纷、建设工程合同纠纷、劳动争议、交通事故责任纠纷、离婚纠纷、机动车交通事故责任纠纷
文书类型	判决书/裁定书/调解书	判决书（默认）
地区	省份名称	广东省、北京市、上海市、浙江省、江苏省、四川省、山东省
年份	裁判年份（4位数字）	2025、2026
审判程序	一审/二审/再审	一审,二审（默认，同时搜两种）

---

运行方式

# 默认：测试模式，仅抓前10条
python wenshu_search.py

# 测试模式（显式）
python wenshu_search.py --test

# 全量模式（抓全部文书）
python wenshu_search.py --full

⚠️ 全量模式建议正式使用时分批运行，避免短时间内大量访问导致账号风控。

---

首次配置向导

首次运行（无 .wenshu_config.json 时）会显示交互式配置：

============================================================
  裁判文书网搜索 v10 — 首次配置向导
============================================================
  提示：配置只输入一次，之后自动记忆（密码除外）
        按 Enter 使用默认值

📋 裁判文书网账号（手机号）[]：
🔑 裁判文书网密码：
🔍 案由（中文）[默认=房屋租赁合同纠纷]：
📄 裁判文书类型 [判决书 / 裁定书 / 调解书，默认=判决书]：
⚖️  审判程序（一审/二审/再审，可用逗号分隔）[默认=一审,二审]：
🌏 地区（中文省份）[默认=广东省]：
📅 裁判年份（4位数字）[默认=当前年份]：

---

Excel 输出字段（13列）

序号 / 案件名称 / 案号 / 判决书内容 / 法院名称 / 法院省份 / 法院地市 / 案件类型 / 审判程序 / 案由 / 裁判年份 / 裁判日期 / 当事人

• 判决书全文直接嵌入 Excel 第4列，已清洗网页噪音
• 全文超32000字自动截断并标注
• 所有列启用自动换行，判决书内容列宽60，其余紧凑

---

合规使用准则

• ✅ 只模拟正常用户操作（登录、点击、翻页）
• ✅ 只读监听，不拦截不篡改任何请求
• ✅ 合理间隔，不高频批量访问
• ❌ 严禁 page.route() 拦截/篡改请求
• ❌ 严禁绕过服务器安全机制
• ⚠️ 账号被封禁风险与访问频率正相关，请勿高频批量抓取

---

文件结构

fanghongyuanwenshu/
├── SKILL.md              ← 本文件
└── scripts/
    └── wenshu_search.py  ← 主脚本（通用版，已脱敏）

---

常见案由代码参考

本技能支持以下常见案由（案由代码自动对应）：

房屋租赁合同纠纷    民间借贷纠纷      买卖合同纠纷
建设工程合同纠纷    劳动争议          交通事故责任纠纷
离婚纠纷            机动车交通事故责任纠纷
服务合同纠纷        租赁合同纠纷

如需其他案由，修改脚本中 S14_CODE_SEARCH 和 S14_CODE_FILTER 对应代码即可。

---

本技能基于裁判文书网公开数据，仅供法律研究与案件参考使用。