DarnShure/my_mcp_server

my_mcp_server

This project provides an end-to-end testable MCP (Model Context Protocol) server environment, including a FastAPI-based chatbot generation stub, an MCP server, and a client stub for testing the generate tool. The main focus is on enabling easy local testing of the chatbot generation workflow.

TLDR: This package is responsible for end-to-end pdf preprocessing for llm ingestion.

It adds several features onto pdfplumber

MCP Server Feedback Mechanism Agentic adaptive pdf parsing. Iterative evolutionary framework.

Project Structure

my_mcp_server/
├── [`chatbot_generation_stub.py`](chatbot_generation_stub.py )
├── [`client_stub.py`](client_stub.py )
├── common/
│   └── [`common/config.py`](common/config.py )
├── config/
│   └── [`config/custom_config.toml`](config/custom_config.toml )
├── [`main.py`](main.py )
├── [`run_all_test.sh`](run_all_test.sh )
├── [`run_inspector.sh`](run_inspector.sh )
├── [`run_mcp_server.sh`](run_mcp_server.sh )
├── [`server.py`](server.py )
├── [`pyproject.toml`](pyproject.toml )
└── ...

Quick Start: End-to-End Testing

The recommended way to test the full pipeline, including the generate tool, is to use the script. This script will:

1. Read the chatbot API port from .
2. Wait for the chatbot API to be available at the configured port.
3. Start the chatbot API stub ().
4. Start the MCP server ().
5. Start the MCP inspector ().
6. Optionally, run the client stub () to test the generate tool.
7. Clean up all background processes after testing.

How to Run

1. Install dependencies
   Make sure you have Python 3.10+ and Node.js installed.
   Install Python dependencies:

   pip install -r requirements.txt
   

   Or, if using Poetry or PDM, install via your preferred tool.

2. Run the test script
   From the project root, execute:

   bash run_all_test.sh
   

   The script will:

   • Wait for the chatbot API to be available at http://127.0.0.1:<port>/ (port from config).
   • Start all required services.
   • Prompt you to continue after setup.
   • (Optionally) Run the client stub to test the tool.

3. Test the tool
   The client stub () connects to the MCP server and calls the tool, which in turn calls the chatbot API stub and prints the results.

Configuration

• The chatbot API port and other settings are configured in .
• The MCP server reads its configuration from this file via .

Components

• : FastAPI app simulating the chatbot generation endpoint.
• : MCP server exposing the tool.
• : Example client that connects to the MCP server and tests the tool.
• : Orchestrates the full end-to-end test environment.

Notes

• The script assumes all commands (uv, npx, etc.) are available in your PATH.
• For custom configuration, edit .

---

For more details on each component, see