legal-search

AI legal database retrieval assistant

A full-process legal retrieval automation tool for lawyers, judge assistants, and legal affairs personnel. With one sentence of input, it automatically completes legal and regulatory retrieval → case verification → similar case analysis → report generation.

What it can do

📜 Legal and regulatory retrieval: Describe a legal issue in natural language, and automatically retrieve relevant legal provisions (including timeliness, level of authority, and original text of legal articles)
🔍 Case verification: Retrieve judgment documents by case number or keywords, and obtain the original text of authoritative precedents
📊 Similar case analysis: Find similar cases based on semantic retrieval, and compare judgment points, sentencing differences, and legal application
📄 Visualized report: Generate an interactive HTML report, including original legal texts, case summaries, similar case comparison tables, and extraction of judicial viewpoints

Data sources

This Skill supports dual data sources:

Data source	Coverage	Access method	Price
PKULaw MCP	5 million+ regulations, 160 million+ cases, 42.5 million+ enforcement documents	MCP Streamable HTTP	Free for 1 month for new users, pay-as-you-go
Yuandian Open Platform	4 million+ regulations, 150 million+ judgment documents, enterprise litigation involvement	MCP / REST API	50,000 points per month (April-July 2026)

Recommended: Prioritize the Yuandian Open Platform (sufficient free quota during the promotional period), with PKULaw as a supplement (more authoritative data).

Quick start

Method 1: MCP direct connection (recommended, most complete data)

1. Configure Yuandian MCP (free quota, priority)

# 1. 访问 https://open.chineselaw.com/ 注册获取 API Key
# 2. 将以下配置添加到 WorkBuddy MCP 配置

# 配置示例：
{
  "mcpServers": {
    "yuandian-mcp-server": {
      "command": "npx",
      "args": ["-y", "yuandian-mcp-server"],
      "env": {
        "YUANDIAN_API_KEY": "你的_api_key"
      }
    }
  }
}

2. Configure PKULaw MCP (authoritative supplement)

# 1. 访问 https://mcp.pkulaw.com/ 注册获取 Access Token
# 2. 在控制台购买需要的MCP服务
# 3. 将以下配置添加到 WorkBuddy MCP 配置

{
  "mcpServers": {
    "pkulaw-law-search": {
      "type": "streamable-http",
      "streamable": true,
      "url": "https://apim-gateway.pkulaw.com/mcp-law-search-service",
      "headers": {
        "Authorization": "Bearer 你的_token"
      }
    }
  }
}

After configuration, you need to click "Trust" on the WorkBuddy connector management page to enable it.

Method 2: Experience demo mode (no configuration required)

No API Key is required; use the built-in legal knowledge base to generate a demo report:

python ~/.workbuddy/skills/legal-search/scripts/demo.py "劳动合同纠纷"

Use cases

Scenario 1: Legal and regulatory retrieval

The user inputs a legal issue in natural language, and relevant legal provisions are automatically retrieved.

用户：查一下关于"网络侵权责任认定"的法律规定

The Skill automatically executes:

Call Yuandian yuandian_law_vector_search semantic retrieval
Filter by timeliness (prioritize currently effective)
Return the original text of legal provisions + regulation name + level of authority
Organize into structured results

Scenario 2: Judicial case verification

The user provides a case number or case keywords, and relevant judgment documents are retrieved.

用户：查案号"(2024)京0105民初12345号"的判决情况

The Skill automatically executes:

Call PKULaw "Case Number Identification and Tracing" to precisely locate the case
Obtain the original text of the judgment document
Extract the trial court, judgment date, and judgment result
Output a case summary

Scenario 3: Similar case analysis report

The user describes the case situation, and a similar case analysis report is automatically generated.

用户：分析"外卖骑手与平台之间的劳动关系认定"相关类案

The Skill automatically executes:

Legal and regulatory retrieval: Retrieve relevant provisions of the Labor Contract Law
Case retrieval: Semantically retrieve relevant judgment documents
Similar case analysis: extract key judicial points, compare views of different courts
Report generation: output an HTML report containing the original legal provisions, a case comparison table, and a distillation of judicial views

Workflow

用户输入法律问题/案号/案情描述
    │
    ├─ 阶段1：法规检索
    │   ├─ 元典语义检索 → 相关法条
    │   ├─ 北大法宝关键词检索 → 补充法规
    │   └─ 法条溯源验证 → 确认有效性
    │
    ├─ 阶段2：案例查证
    │   ├─ 有案号 → 北大法宝案号溯源
    │   ├─ 无案号 → 元典/北大法宝语义案例检索
    │   └─ 获取裁判文书摘要
    │
    ├─ 阶段3：类案分析
    │   ├─ 相似案例聚类（按案由/法院层级/裁判年份）
    │   ├─ 裁判观点提炼与对比
    │   └─ 法律适用差异分析
    │
    ├─ 阶段4：报告生成
    │   ├─ 合成结构化数据
    │   ├─ 渲染HTML可视化报告
    │   └─ 输出报告文件
    │
    └─ 输出：legal_report_{timestamp}.html

Agent Execution Instructions

When a user triggers a legal retrieval request, the Agent should execute the following steps:

Step 1: Parse user intent

Extract from user input:

Retrieval type: legal provision retrieval / case verification / similar case analysis / comprehensive
Query keywords: description of the core legal issue
Known information: case number, legal provision name, court, time range, etc.
Output format preference: brief result / detailed report

Step 2: Data retrieval (select based on intent)

Pure legal provision retrieval scenario：

优先使用元典 yuandian_law_vector_search
备选使用北大法宝 "检索法律法规-语义"

Case verification scenario：

有案号 → 北大法宝 "案号识别与溯源" → "检索司法案例-关键词"
无案号 → 元典案例语义检索 → 北大法宝 "检索司法案例-语义"

Similar case analysis scenario：

1. 先做法规检索（了解法律依据）
2. 再做案例检索（多组关键词，覆盖不同角度）
3. 最后做类案聚类分析

Step 3: AI analysis and integration

Perform the following on the retrieval results:

Legal provision analysis: arrange by level of legal effect, mark timeliness, and extract the provisions most relevant to the case
Case summary: extract the trial court, case number, judgment date, disputed issues, and judgment result
Similar case comparison: create a comparison table of judicial views, marking consensus and disagreements
Risk alerts: based on similar case results, indicate litigation risks and key evidence requirements

Step 4: Generate HTML report

Call scripts/report_generator.py to generate an interactive HTML report:

python ~/.workbuddy/skills/legal-search/scripts/report_generator.py \
  --input <search_results.json> \
  --output <report.html> \
  --title "报告标题"

The report includes:

Retrieval overview: retrieval conditions, data sources, result statistics
Laws and regulations: original text of relevant legal provisions, level of legal effect, timeliness labels
Case list: case summary cards (case number, court, judgment date, result)
Similar case comparison: comparison table of judicial views + visualization charts
Analysis recommendations: legal analysis recommendations based on similar cases

Technical architecture

用户输入
    ↓
[意图解析] → 识别检索类型与关键信息
    ↓
[MCP调用层] → 元典/北大法宝 API
    ↓
[数据整合] → 结构化法律数据
    ↓
[AI分析] → 类案聚类 + 观点提炼
    ↓
[报告生成] → scripts/report_generator.py
    ↓
HTML交互式报告

Notes

🔑 MCP configuration: first-time use requires registration and configuration of at least one data source (Yuandian free quota is sufficient)
📊 Number of results returned: case retrieval is recommended to be kept within 20-50 items to avoid an overly long report
⚖️ Timeliness: legal provision retrieval filters out invalid provisions by default and prioritizes currently effective provisions
💰 Cost control: Yuandian provides 50,000 free points per month; PKULaw is free for new users in the first month
🏛️ Data limitations: updates on China Judgments Online have narrowed since 2025; cross-validation with multiple data sources is recommended
📝 AI analysis: similar case analysis depends on AI model capabilities; it is recommended to refer to the original case texts for final judgment

Demo mode

Experience the full functionality without any API Key:

# 生成演示报告
python ~/.workbuddy/skills/legal-search/scripts/demo.py "外卖骑手劳动关系认定"

# 输出：legal_report_demo.html

The demo report is based on a built-in legal knowledge base and includes authentic original legal provisions and simulated case data.

Changelog

v1.0.0 (2026-06-20): Initial version
- Supports Yuandian Open Platform MCP integration (semantic retrieval of laws and regulations + case retrieval)
- Supports PKULAW MCP integration (statute/case/case number tracing)
- Three major scenarios: statute search, case verification, similar-case analysis reports
- Interactive HTML visualization report output
- Built-in demo mode (experience it without an API Key)

Related Skills

contract-review: Contract Review Assistant (works with this skill to review contracts after searching relevant statutes)
tender-analyzer: Bid Document Analysis (search for the legal basis of bidding and tendering regulations)
fifteen-five-plan: 15th Five-Year Plan Interpretation (policy and regulation analysis)

Author: WorkBuddy Skill Creator Feedback: If you encounter any issues or have suggestions for improvement, please provide feedback in WorkBuddy

About this skill

What it can do

Data sources

Quick start

Method 1: MCP direct connection (recommended, most complete data)

1. Configure Yuandian MCP (free quota, priority)

2. Configure PKULaw MCP (authoritative supplement)

Method 2: Experience demo mode (no configuration required)

Use cases

Scenario 1: Legal and regulatory retrieval

Scenario 2: Judicial case verification

Scenario 3: Similar case analysis report

Workflow

Agent Execution Instructions

Step 1: Parse user intent

Step 2: Data retrieval (select based on intent)

Step 3: AI analysis and integration

Step 4: Generate HTML report

Technical architecture

Notes

Demo mode

Changelog

Related Skills