rag-mcp-server

Tsarri/rag-mcp-server

3.2

If you are the rightful owner of rag-mcp-server and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to dayong@mcphub.com.

A comprehensive multi-agent Retrieval-Augmented Generation (RAG) system built on the Model Context Protocol (MCP), featuring specialized AI microagents for legal document processing, deadline extraction, and strategic analytics.

Tools
9
Resources
0
Prompts
0

Multi-Agent RAG MCP Server

A comprehensive multi-agent Retrieval-Augmented Generation (RAG) system built on the Model Context Protocol (MCP), featuring specialized AI microagents for legal document processing, deadline extraction, and strategic analytics.

🎯 Overview

This project implements an interconnected agentic ecosystem using MCP servers as the foundation for coordinating specialized AI agents. The system is designed for legal tech applications, particularly document intelligence and deadline management.

✨ Features

  • Multi-Agent Architecture: Three specialized agents working in coordination
  • Vector Storage: Supabase with pgvector for semantic search
  • MCP Integration: Seamless integration with Claude Desktop
  • Legal Document Processing: Specialized for Spanish legal notifications
  • Strategic Analytics: Business intelligence and context analysis
  • Zero-Input Strategy: 75% automation, 25% strategic oversight

🤖 Agents

1. Deadline Agent

Extracts and manages deadlines from Spanish legal documents with high accuracy.

Capabilities:

  • Spanish legal text processing
  • Deadline extraction and categorization
  • Automated deadline tracking
  • Legal notification parsing

2. Document Classification Agent

Automatically categorizes and classifies legal documents.

Capabilities:

  • Multi-class document classification
  • Metadata extraction
  • Automated tagging
  • Document type recognition

3. SmartContext Analytics Agent

Provides strategic business intelligence and contextual analysis.

Capabilities:

  • Strategic analytics
  • Business context extraction
  • Cross-document insights
  • Trend analysis

🏗️ Architecture

rag-mcp-server/
├── src/
│   ├── server.py              # Main MCP server
│   ├── agents/
│   │   ├── deadline_agent.py
│   │   ├── document_agent.py
│   │   └── smartcontext_agent.py
│   └── data_sources/
├── database/
│   └── schema.sql             # Database schema
├── docs/                      # Documentation
├── config/                    # Configuration files
├── data/                      # Data storage
└── tests/                     # Test files

🚀 Quick Start

Prerequisites

  • Python 3.10+
  • Supabase account
  • Claude Desktop (for MCP integration)
  • PostgreSQL with pgvector extension

Installation

  1. Clone the repository
git clone https://github.com/yourusername/rag-mcp-server.git
cd rag-mcp-server
  1. Create virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
  1. Install dependencies
pip install -r requirements.txt
  1. Configure environment
cp .env.example .env
# Edit .env with your credentials
  1. Initialize database
# Run the database schema (see docs for details)
psql -h your-supabase-host -U postgres -d your-database -f database/schema.sql
  1. Configure Claude Desktop Edit your Claude Desktop config file (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):
{
  "mcpServers": {
    "rag-server": {
      "command": "python",
      "args": ["/Users/yourusername/rag-mcp-server/src/server.py"],
      "env": {
        "SUPABASE_URL": "your_supabase_url",
        "SUPABASE_KEY": "your_supabase_key"
      }
    }
  }
}
  1. Restart Claude Desktop

🛠️ Usage

The system can be used in two ways:

1. MCP Server (Claude Desktop Integration)

Once configured, the agents are available through Claude Desktop with the following tools:

Deadline Agent Tools

  • extract_deadlines - Extract deadlines from legal documents
  • list_deadlines - List all tracked deadlines
  • search_deadlines - Search deadlines by criteria

Document Agent Tools

  • classify_document - Classify document type
  • index_document - Add document to vector store
  • search_documents - Semantic document search

SmartContext Agent Tools

  • analyze_context - Strategic context analysis
  • extract_insights - Business intelligence extraction
  • trend_analysis - Cross-document trend analysis

2. REST API Server (Frontend Integration)

The system also provides a FastAPI REST API for frontend applications:

# Run REST API server (for frontend)
python src/api_server.py

The API server runs on http://localhost:8000 with interactive documentation at http://localhost:8000/docs.

Key Features:

  • Client Management - Create and manage client records
  • Document Upload - Upload documents with automatic processing
  • Data Retrieval - Query documents, deadlines, and analyses per client
  • CORS Enabled - Ready for frontend integration

API Endpoints:

Client Management:

  • POST /api/clients - Create new client
  • GET /api/clients - List all clients
  • GET /api/clients/{client_id} - Get client details
  • PUT /api/clients/{client_id} - Update client
  • DELETE /api/clients/{client_id} - Delete client (soft delete)

Document Operations:

  • POST /api/clients/{client_id}/documents - Upload and process document
  • GET /api/clients/{client_id}/documents - List client's documents
  • GET /api/clients/{client_id}/documents/stats - Document statistics

Deadline Management:

  • GET /api/clients/{client_id}/deadlines - Get client's deadlines
  • GET /api/clients/{client_id}/deadlines/stats - Deadline statistics

Strategic Analysis:

  • GET /api/clients/{client_id}/analysis - Get strategic analyses

Running Both Servers:

# Run MCP server for Claude Desktop (existing functionality)
python src/server.py

# Run REST API server for frontend (new functionality)
python src/api_server.py

Both servers can run independently and use the same database.

📊 Database Schema

The system uses the following main tables:

  • clients - Client information and management
  • documents - Document metadata and classification
  • deadline_extractions - Deadline extraction operations
  • deadlines - Extracted deadline tracking
  • analyses - Strategic insights and analytics

Client Isolation: All documents, deadlines, and analyses are associated with specific clients via client_id, enabling proper data isolation and multi-tenant support.

See database/schema.sql for complete schema details.

📚 Documentation

Comprehensive documentation is available in the docs/ folder:

  • Quick Start Guide - 30-minute setup from scratch
  • Architecture Guide - Complete system design and patterns
  • Troubleshooting Guide - Common issues and solutions
  • API Reference - Tool definitions and usage

🔒 Security

This system implements three-layered security:

  1. Authentication - User identity verification
  2. Authorization - Access control and permissions
  3. Encryption - Zero-knowledge encryption for sensitive data

Never commit your .env file - it contains sensitive credentials.

🧪 Testing

Run the test suite:

pytest tests/

Test individual agents:

python test_deadline_extraction.py

🤝 Contributing

This is a personal project, but suggestions and feedback are welcome! Please open an issue to discuss proposed changes.

📝 License

[Add your license here]

🙏 Acknowledgments

Built with:

📞 Contact

[Add your contact information]

Status: Production-Ready
Version: 1.0
Last Updated: November 2024