📝 Task Manager MCP Server

📖 Overview

The Task Manager MCP Server is a Spring Boot application that exposes a Model Context Protocol (MCP) interface.
It enables AI agents like Claude, GPT-4, etc., to interact with a shared PostgreSQL Task Management Database.

Key features:

• Schema inspection (/mcp/schema/tasks)
• Bulk task insertion (/mcp/tasks)
• Summary analytics (/mcp/tasks/summary)
• Agent-readable help (/mcp/help)

🧩 Architecture

Mermaid Diagram (GitHub-ready)

flowchart TD
    A["AI Agent (Claude / GPT-4 / etc.)"] <--> B["MCP Server (Spring Boot)"]
    B <--> C["PostgreSQL Database"]

    A --> B
    B --> C
    B --> D["PostgreSQL Database"]
    C --> D 

Diagram Explanation

• AI Agent: The AI agent that interacts with the MCP server.
• MCP Server: The Spring Boot application that exposes the MCP interface.
• PostgreSQL Database: The shared database where tasks are stored. er connects to the PostgreSQL database.

Technology Stack

• Spring Boot 3.2: The backend framework for building the MCP server.
• PostgreSQL: The database used for storing tasks.
• MCP: The Model Context Protocol used for communication between the AI agent and the MCP server.
• Java 17+: The programming language used for the backend code.
• Spring Data JPA: The ORM (Object-Relational Mapping) library used for database interactions.
• Springdoc OpenAPI/Swagger: The documentation framework used for generating API documentation.

🚀 How to Run

Clone the repository:

• git clone https://github.com/your-org/taskmanager-mcp-server.git
• cd taskmanager-mcp-server
• Update application.properties (or application.yml) with your local DB:

spring.datasource.url=jdbc:postgresql://localhost:5432/taskdb
spring.datasource.username=task_user
spring.datasource.password=task_pass
spring.jpa.hibernate.ddl-auto=update

Run the application:

• mvn spring-boot:run
• The server starts at: 👉 http://localhost:8080

Swagger UI (interactive API docs):

👉 http://localhost:8080/swagger-ui/index.html

🧠 AI Agent (Claude) Integration

• Claude (or another MCP-enabled AI agent) connects via a config file.
• Example claude_desktop_config.json:

{
 "mcpServers": {
   "task-manager": {
     "command": "node",
     "args": [
       "C:\\Users\\your-username\\IdeaProjects\\mcp\\mcp\\mcp-proxy.js"
     ]
   }
 }
}

Notes for other machines:

• Replace C:\Users\your-username\IdeaProjects\mcp\mcp\mcp-proxy.js with the absolute path to your mcp-proxy.js.
• On Linux/Mac, the path might look like:

"/home/username/projects/mcp/mcp-proxy.js"

• Make sure the path is correct and accessible by the AI agent.

💬 Example Prompt to Claude

Please inspect the task schema at /mcp/schema/tasks.  
Then generate and insert 1000 diverse tasks with random statuses, titles, and due dates using the /mcp/tasks endpoint.  
Finally, verify insertion using /mcp/tasks/summary.

✅ Expected Results

• /mcp/schema/tasks → Returns schema of Task
• /mcp/tasks → Inserts ~1000 tasks
• /mcp/tasks/summary → Returns analytics:

{
  "totalTasks": 1000,
  "statusDistribution": {
    "TODO": 250,
    "IN_PROGRESS": 200,
    "DONE": 350,
    "CANCELLED": 100,
    "ON_HOLD": 100
  },
  "priorityDistribution": {
    "LOW": 200,
    "MEDIUM": 400,
    "HIGH": 300,
    "URGENT": 100
  }
}

📚 Deliverables

✅ Source code (Spring Boot + MCP server)

✅ PostgreSQL config

✅ README.md (this file)

✅ Example Claude prompts + results

✅ Swagger UI (interactive API docs)