ai-code-helper

This MCP server provides an AI-assisted coding mentor and interview preparation assistant built with LangChain4j and the Tongyi Qianwen model. It combines a backend Spring Boot service with a Vue.js frontend and exposes a set of tools for document search, retrieval augmented generation (RAG), and safe input handling to help users learn programming and prepare for interviews. The MCP protocol enables you to manage contexts and model calls in a unified way, so you can plug this server into your broader AI tooling and pipelines. Typical usage involves running the backend to host the API and optionally serving the frontend for a complete UI experience. The server coordinates between the LangChain4j layer, the MCP context protocol, and the model endpoints to deliver streaming responses and structured outputs.

Key capabilities include:

• Conversational AI for programming guidance, resume and interview coaching, and code examples.
• RAG-based retrieval from local knowledge bases to surface relevant information.
• Tool integration for searching interview questions and web-based resources.
• Streaming response support for a real-time typing-like user experience.
• Safety guards for input to protect against sensitive content.

To use the MCP server, first start the backend (Spring Boot) and ensure your Tongyi Qianwen API key and any required big-model API keys are configured. Then connect your MCP client or frontend to the backend API endpoints to begin interacting with the AI assistant. The frontend (Vue.js) provides a chat UI and real-time streaming, while the backend exposes RESTful API endpoints and MCP-integrated capabilities.

Prerequisites:

• Java 21+ (JDK)
• Maven 3.6+ (for building and running the Spring Boot app)
• Access to Tongyi Qianwen API and any Big Model API keys

Install steps:

1. Clone the repository: git clone <repository-url> cd ai-code-helper

2. Set up API keys (example for application.yml or environment variables):

   • TH_API_KEY=your Tongyi Qianwen API key
   • BIG_MODEL_API_KEY=your Big Model API key
   • If you expose via environment variables, ensure Spring Boot reads them or configure in application.yml accordingly.

3. Build and run the backend (Spring Boot): mvn spring-boot:run

4. Start the frontend (optional, if a frontend is provided): cd ai-code-helper-frontend npm install npm run dev

5. Access the application: Backend API: http://localhost:8080/api Frontend (if used): http://localhost:5173

Tips and common considerations:

• Ensure your API keys are securely stored and not committed to version control.
• The MCP integration relies on proper model and context management; if you see context-related errors, verify your MCP settings and ensure the backend is up-to-date with the protocol version.
• For local development, you may map different ports if 8080 or 5173 are occupied.
• If you enable streaming, ensure your client supports receiving and rendering incremental messages.
• The server can be extended with additional tools (e.g., more interview question sources or custom knowledge bases) by configuring LangChain4j pipelines and MCP tool integrations.
• Check logs for any connectivity issues with Tongyi Qianwen or the big-model API endpoints and update keys as needed.