Dataset Creating AI Assistant

A Spring Boot application that serves as an AI-powered assistant to generate training datasets using MCP (Model Context Protocol) for web content extraction and email notifications.

Overview

This application provides:

• Content extraction from target websites
• Generation of training datasets in JSON format
• Optional email notifications upon dataset completion
• Conversation history with memory management
• MCP tools integration for web scraping and mailing

Features

• AI Chat Interface: RESTful API for conversational AI interactions
• Dataset Generation: Automated creation of training datasets from any website
• Web Content Extraction: MCP fetch tools for different content types (markdown, HTML, raw text)
• Memory Management: Conversation history with configurable message window
• French Language Support: All interactions and outputs in French
• Email Notifications: Automatic emails when dataset generation is completed (via custom MCP SMTP server)
• Transport Support: STDIO for fetch server and SSE for email server

Technology Stack

• Java 21
• Spring Boot 3.5.3
• Spring AI 1.0.0
• Groq AI API (Llama models)
• MCP SDK 0.11.0
• MCP Client for Spring AI

Prerequisites

• Java 21 or higher
• Maven 3.6+
• Node.js (for MCP servers)
• Access to Groq API

MCP Servers Setup

Web Fetch Server (STDIO Transport)

Use the TypeScript-based MCP fetch server:

git clone https://github.com/tatn/mcp-server-fetch-typescript.git
# Build and configure path in mcp-servers.json

Custom MCP Email Server (SSE Transport) Use the Java MCP SMTP server:

git clone https://github.com/RayenMalouche/Java-MCP-Server-For-SMTP-Mailing.git
# Follow setup instructions in the repository and configure SMTP settings
# Start server (typically runs on localhost:45450)*

Configuration

Application Properties Example

spring.application.name=Dataset-MCP-Assistant
server.port=8072

# Groq AI Configuration
spring.ai.openai.api-key=your_groq_api_key
spring.ai.openai.base-url=https://api.groq.com/openai
spring.ai.openai.chat.options.model=llama-3.1-8b-instant

# MCP STDIO and SSE Configuration
spring.ai.mcp.client.stdio.servers-configuration=classpath:mcp-servers.json
spring.ai.mcp.client.sse.connections.smtp-email-server.url=http://localhost:45450/sse
spring.ai.mcp.client.enabled=true
spring.ai.mcp.client.toolcallback.enabled=true
spring.ai.mcp.client.tool.name-prefix-enabled=false

# Logging
logging.level.org.springframework.ai=DEBUG
logging.level.io.modelcontextprotocol=TRACE

MCP Servers Configuration (mcp-servers.json)

{
  "mcpServers": {
    "mcp-server-fetch-typescript": {
      "command": "node",
      "args": ["path/to/mcp-server-fetch-typescript/build/index.js"]
    }
  }
}

Installation & Setup

Clone the repository:

git clone https://github.com/RayenMalouche/MCP-Client-Dataset-Creating-Assisstant.git
cd MCP-Client-Dataset-Creating-Assisstant

Configure API keys:

• Update application.properties with your Groq API key
• Configure MCP server paths as shown above

Setup MCP Servers:

• Build and configure the fetch server
• Optionally set up the email server for notifications

Build and run the Spring Boot application:

mvn clean install
mvn spring-boot:run

API Endpoints

Chat Interface

GET /assistant/prompt?userId={userId}&message={message}

• userId: Unique identifier for conversation context
• message: User query in French

Conversation Management

GET /assistant/{userId}/history
DELETE /assistant/{userId}/history

MCP Tools Available

Web Content Extraction (STDIO)

• get_raw_text: Extract raw text from URLs
• get_rendered_html: Get complete HTML after JavaScript execution
• get_markdown: Convert content to structured Markdown
• get_markdown_summary: Extract main content as clean Markdown

Email Notifications (SSE)

• send-email: Send SMTP email notifications via custom MCP server
  Parameters: to, subject, body, cc (optional), bcc (optional)

Dataset Generation Format

Generated datasets follow the format:

{
  "Input": "- Contexte : [extracted context]. - Question : [specific question].",
  "Output": "[Response based on extracted content]"
}

Monitoring & Logging

• Logs: ./ai-agent-visitor.log
• Debug level logging for Spring AI and MCP components
• Actuator endpoints available for health monitoring

Error Handling

• Graceful error messages for invalid URLs or missing data
• Email sending failures are logged but do not stop dataset generation

Contributing

1. Fork the repository
2. Create a feature branch
3. Make changes and test with both MCP servers (if using email features)
4. Submit a pull request

Support

For issues related to:

• Application functionality: Create an issue in this repository
• MCP Email Server: See Java-MCP-Server-For-SMTP-Mailing