Nutrient Document Engine MCP Server

# Document Chat A chat interface for interacting with documents through the Document Engine MCP server using LangGraph and Next.js. ## Overview Document Chat is a monorepo application that provides a conversational interface for interacting with documents. It consists of two main components: - **Agents**: A LangGraph-powered agent that connects to the Document Engine MCP server to provide document-related functionality - **Web**: A Next.js web application that provides a user interface for chatting with the agent ## Prerequisites - Node.js v20 or later - pnpm v10 or later - Document Engine MCP server running (locally or remotely) - API keys for the required services (see Environment Variables) ## Environment Variables Create a `.env` file in the root directory with the following variables: ``` # Required API keys OPENAI_API_KEY="" # Required for the LLM MCP_SERVER_URL="http://localhost:5100/mcp" # URL to your Document Engine MCP server ``` ## Installation ```bash # Install dependencies pnpm install ``` ## Development ```bash # Start both the web and agents applications in development mode pnpm dev # Start only the web application pnpm turbo dev --filter=web # Start only the agents application pnpm turbo dev --filter=agents ``` The web application will be available at http://localhost:3000 and the LangGraph agent will be available at http://localhost:2024. ## Building for Production ```bash # Build all applications pnpm build # Build only the web application pnpm turbo build --filter=web # Build only the agents application pnpm turbo build --filter=agents ``` ## Architecture ### Agents Application The agents application uses LangGraph to create a conversational agent that can interact with documents through the Document Engine MCP server. The agent uses: - OpenAI's GPT-4.1-mini model (configurable) - LangGraph for agent orchestration - Document Engine MCP client for document-related tools ### Web Application The web application provides a user interface for chatting with the agent. It uses: - Next.js for the web framework - React for the UI - TailwindCSS for styling - LangGraph SDK for connecting to the agent ## Customization You can customize the agent by modifying the following files: - `apps/agents/src/react-agent/graph.ts`: The main agent graph - `apps/agents/src/react-agent/configuration.ts`: Agent configuration ## License See the LICENSE file in the root directory.