Mcpservercrm

🚀 AI Server MCP for CRM

This server implements the Model Context Protocol (MCP) to provide a secure data processing layer between CRM and Large Language Models (LLMs).

✨ Features

1. Data Security

• Automatically anonymize personal information.
• Use a token system for data tracking without exposing sensitive information.
• Validate data before processing with the LLM.

2. Chatbot Integration

• Manage conversation context.
• Build a Q&A system and provide feedback.
• Continuously evaluate the quality of answers.

3. Lead Analysis

• Automatically assess potential value.
• Track interaction records.
• Provide engagement metrics.
• All operations are performed without exposing personal information.

📦 Installation

Step 1: Create an .env file with the following variables

SECRET_KEY="Your secret key"
DATABASE_URL="Your Supabase PostgreSQL address"
SUPABASE_URL="Your Supabase URL"
SUPABASE_KEY="Your Supabase key"
SUPABASE_JWT_SECRET="Your JWT key"
DEFAULT_LLM_PROVIDER="openai"
DEFAULT_LLM_MODEL="gpt-4"
LLM_API_KEY="Your OpenAI API key"

Step 2: Install dependencies

pip install -r requirements.txt

Step 3: Start the server

uvicorn app.main:app --reload

📚 Documentation

Token and Authentication

• POST /api/v1/tokens/generate: Generate an access token.

Message Processing

• POST /api/v1/messages/sanitize: Sanitize messages for processing.
• POST /api/v1/chatbot/context: Manage the chatbot's context.
• POST /api/v1/qa-pairs: Create Q&A pairs.
• POST /api/v1/evaluate: Evaluate information using the LLM.

Analysis Functions

• POST /api/v1/analyze-lead: Securely analyze leads.
• GET /api/v1/lead-metrics/{lead_id}: Get historical engagement metrics.

🔧 Technical Details

1. Anonymization Layer

• Clean personal information.
• Generate anonymous tokens.
• Validate content legitimacy.

2. Context Layer

• Manage conversation context.
• Build a Q&A system.
• Save sanitized chat records.

3. Analysis Layer

• Evaluate leads.
• Record interaction data.
• Provide aggregated reports.

Database Integration

This server uses Supabase as the backend. The main tables include:

• mensajes_sanitizados: Store sanitized message versions.
• contexto_conversacional: Maintain conversation context.
• pii_tokens: Manage anonymous tokens.
• chatbot_contextos: Save chatbot configurations.
• qa_pares: Store Q&A pairs for training.
• evaluaciones_llm: Record LLM evaluation results.

Continuous Improvement

The system includes:

• Automated evaluation of answer quality.
• Providing feedback to optimize performance.
• Monitoring LLM performance metrics.