MCP_HR_HELPER

# MCP HR Helper (v1) An AI-integrated Leave Management System built using the **Model Context Protocol (MCP)**. This project demonstrates how backend tools can be exposed to LLMs (like Claude) to perform real-world operations such as employee management and leave workflows. --- ## Features ### Employee Management * Add new employees * Delete employees * Search employees by name * View all employee details * Update leave balance ### Leave Management * Apply leave * Approve / Reject leave * Cancel leave * Track leave status * View employee leave history ### Analytics * Leave summary (total, approved, pending, rejected) * Low balance employee detection ### AI Integration * MCP tools callable via Claude * Prompt-based leave email generation * Resource-based employee lookup (`employee://{id}`) --- ## Tech Stack * Python * MCP (Model Context Protocol) * FastMCP Server * JSON (file-based persistence) --- ## 📁 Project Structure ``` MCP_HR_HELPER/ │── main.py │── employees.json │── leaves.json │── .gitignore │── README.md ``` --- ## How It Works * Data is stored in **JSON files** * MCP server exposes tools via `stdio` * Claude interacts with tools dynamically * All updates persist across restarts --- ## Setup Instructions ### 1. Clone Repository ``` git clone https://github.com/your-username/MCP_HR_HELPER.git cd MCP_HR_HELPER ``` ### 2. Create Virtual Environment ``` python -m venv .venv .venv\Scripts\activate ``` ### 3. Install Dependencies ``` pip install mcp ``` ### 4. Run Server ``` python main.py ``` --- ## MCP Configuration (Claude) Add this in: ``` %APPDATA%\Claude\claude_desktop_config.json ``` ``` { "mcpServers": { "leave-manager": { "command": "C:\\path\\to\\.venv\\Scripts\\python.exe", "args": ["C:\\path\\to\\main.py"] } } } ``` --- ## Example Queries * "Show all employees" * "Apply leave for employee 101 for 2 days" * "Approve leave with id 1" * "Show leave summary" * "Generate leave email for Subhasish for 3 days" --- ## Limitations (v1) * No authentication * No database (JSON-based storage) * No concurrency handling * Single-user system --- ## Future Improvements * SQLite / PostgreSQL integration * Role-based access (Admin / Employee) * REST API + Frontend dashboard * Multi-agent HR assistant * Deployment (Docker / Cloud) --- ## Author **Subhasish** B.Tech | AI & Data Science ---