Skip to content

CLAUDE.md

Latest commit

 

History

History
117 lines (83 loc) · 3.75 KB

CLAUDE.md

File metadata and controls

117 lines (83 loc) · 3.75 KB

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Project Overview

This is a FastAPI-based API server for accessing Japanese legal information (laws and court precedents). The project uses uv for package management and is designed with future integration of Agentic RAG and Claude Agent SDK in mind.

Current Status: Phase 1 - Foundation building (project initialization stage)

Development Setup

This project uses uv as the package manager. Setup commands:

# Create virtual environment
uv venv

# Activate virtual environment
source .venv/bin/activate  # Linux/Mac
.venv\Scripts\activate     # Windows

# Install dependencies
uv pip install -e .

Architecture Overview

Planned Structure

The project is designed around three main data layers:

  1. Legal Data Layer (app/services/egov_client.py)

    • Integrates with e-gov API (https://elaws.e-gov.go.jp/) for Japanese law data
    • Handles law searches, retrieval, and amendment history

  2. Case Law Layer (app/services/case_scraper.py)

  3. AI Integration Layer (app/services/rag_service.py)

    • Agentic RAG for efficient search across legal data
    • Claude Agent SDK integration for custom tools
    • MCP server implementation for tool provision

API Design

The API will follow RESTful patterns with versioning:

Laws endpoints:

  • GET /api/v1/laws/search - Search laws
  • GET /api/v1/laws/{law_id} - Get specific law
  • GET /api/v1/laws/{law_id}/history - Get amendment history

Cases endpoints:

  • GET /api/v1/cases/search - Search court precedents
  • GET /api/v1/cases/{case_id} - Get specific case

Claude Agent SDK Integration

The project will expose legal search functionality as MCP tools:

@tool("search_law", "日本の法令を検索", {"query": str})
async def search_law(args):
    # e-gov API を使用した法令検索
    results = await egov_client.search(args["query"])
    return {"content": [{"type": "text", "text": results}]}

@tool("search_case", "判例を検索", {"keywords": str})
async def search_case(args):
    # 判例データベースでの検索
    results = await case_service.search(args["keywords"])
    return {"content": [{"type": "text", "text": results}]}

Key Technical Decisions

  • Package Manager: uv (for fast, reliable dependency management)
  • API Framework: FastAPI (for async support and automatic OpenAPI documentation)
  • Python Version: 3.10+ (required for modern async features)
  • Claude Integration: Agent SDK with MCP server pattern (not direct API calls)

Data Sources

  1. e-gov Law API: Official source for all Japanese laws, real-time updates
  2. Courts Website: Source for court precedent data
  3. Future considerations: Commercial case databases, open legal data

Development Phases

Phase 1 (Current): Foundation

  • FastAPI project initialization
  • e-gov API client implementation
  • Basic law search endpoints

Phase 2: Case law integration

  • Data source selection for court precedents
  • Scraper/API client implementation
  • Case search endpoints

Phase 3: AI integration

  • Agentic RAG implementation
  • Claude Agent SDK integration
  • MCP server functionality

Phase 4: Advanced features

  • Law-case relationship analysis
  • Natural language legal consultation
  • Automatic legal document summarization

Important Notes

  • All API responses should be in Japanese (法律・判例データは日本語)
  • Legal data accuracy is critical - always verify data sources
  • Consider data update frequency and caching strategies
  • MCP tools should handle both structured queries and natural language