mcp-coding -demo-app

This MCP server demonstrates a Claude AI–driven development workflow integrated through the Model Context Protocol. It provides a structured context management system that preserves project context and conversation history, enabling Claude to assist with Todo List development tasks, code analysis, and repository operations within a Java EE / Quarkus environment. The server handles prompts via a templated prompt system, enforces development guidelines, and coordinates with Claude to generate context-aware responses and actionable recommendations for code changes, commit messages, and repository actions. You can use it to experiment with AI-assisted development prompts, track project state across interactions, and observe how Claude suggests improvements while maintaining project consistency.

Prerequisites:

• Docker installed and running
• Optional: Claude API access (API key) if using an external Claude integration
• Basic knowledge of your environment where the container will run

Installation steps:

1. Pull or build the Docker image for the MCP Coding Server Demo App (if you own the image tag, replace with your own): docker pull lamemind/mcp-coding-server-demo-app:latest

2. Prepare environment variables (example):

   • CLAUDE_API_KEY=your-Claude-api-key

3. Run the MCP server container locally: docker run -i
   -e CLAUDE_API_KEY=$CLAUDE_API_KEY
   -p 8080:8080
   lamemind/mcp-coding-server-demo-app:latest

4. Verify the server is up by accessing the MCP endpoint (for example, http://localhost:8080/mcp or as documented by the image).

If you prefer building locally first (without Docker):

1. Prerequisites:
   • Java 17+ (JDK)
   • Maven
   • MySQL 8+ (if persisting context/history)
2. Build the server components: mvn -f mcp-server/pom.xml clean package
3. Run the server JAR (adjust path if packaged differently): java -jar mcp-server/target/mcp-coding-server-demo-app.jar
4. Ensure database and prompt resources are configured as described in the project docs.

Notes and tips:

• The server relies on prompt templates stored under prompts/; customizations in project prompts.xml and history files in prompts/history/ will influence Claude's responses.
• If you enable Claude integration, supply a valid CLAUDE_API_KEY; without it, the server can still function for local prompt orchestration but will not fetch AI-generated responses from Claude.
• Ensure MySQL (or your chosen backend) is accessible if you enable persistent conversation history or project context storage.
• Validate file system access and version control interactions, as the MCP server enforces context-bound operations and may restrict actions outside the project scope.
• When testing, reset or clear the history for a clean demonstration of context-aware prompting.
• If you encounter port conflicts, modify the docker run command to expose an alternative port (e.g., -p 8081:8080).
• Review prompt templates under prompts/ for how project context and history are composed; customizing these can significantly affect Claude’s guidance.