MCP

DevInsight is an MCP-powered AI assistant designed to analyze, explain, and refactor code locally. It exposes a suite of MCP tools through a Fastify backend and a React-based frontend interface, enabling tasks such as code analysis, natural language explanations, bug detection, refactoring suggestions, test execution summaries, and RAG-driven code search. Users can index their codebases for retrieval-augmented generation (RAG), then interact via REST API endpoints or the chat UI to request analyses, explanations, or refactors. The system relies on an AI provider (OpenAI or Ollama) and a local or remote vector store to perform semantic search over indexed code snippets. In practice, you can index a repository, then ask the chat to analyze a file, explain how a function works, find potential bugs from an error trace, or run tests and summarize results. The API endpoints (/api/mcp/* and /api/rag/*) map to individual MCP tools such as analyze, explain, find-bug, refactor, run-tests, git-history, and rag/index.

Prerequisites:

• Node.js installed (recommended LTS)
• Access to an AI provider (OpenAI API key or Ollama locally)
• Git (optional, for repository operations)

1. Clone the project

• git clone <repository-url>
• cd devinsight

2. Install backend dependencies

cd backend
npm install

3. Install frontend dependencies (optional for full UI)

cd ../frontend
npm install

4. Set up environment variables

cp .env.example .env

Edit .env and add required values:

• OPENAI_API_KEY (if using OpenAI)
• AI_PROVIDER (openai or ollama)
• REPO_PATH (path to repository to analyze)
• VECTOR_DB_PATH (path to vector database)

5) Start the server
```bash
cd backend
npm run dev

6. (Optional) Start the frontend UI

cd frontend
npm run dev

Note: The backend server exposes endpoints like /api/mcp/analyze, /api/mcp/explain, /api/mcp/find-bug, /api/mcp/refactor, /api/mcp/run-tests, and /api/rag/index for indexing and /api/rag/stats for index stats. Ensure your environment variables are correctly set to enable the AI provider and vector store integration.

Tips:

• Ensure OPENAI_API_KEY or Ollama is configured; without a valid API key, MCP tools may return placeholders.
• Before using RAG features, index your repository via POST /api/rag/index with {"repoPath": "./data/your_repo"}.
• For code analysis and refactoring, provide clear references to files or code blocks (e.g., Analyze the code in backend/index.js).
• Git-related features require a valid Git repository at REPO_PATH.
• If using Ollama, ensure the local LLM server is running and accessible as configured in AI_PROVIDER.
• The system uses a vector store (SQLite via better-sqlite3) for fast semantic search; monitor disk usage for large codebases.