Mcp_server_client

What is Mcp_server_client?

A system that provides a chat interface (Streamlit) backed by a NestJS API, integrating multiple MCP (Model Context Protocol) servers for tool execution, RAG (Retrieval-Augmented Generation) with a vector store, and OpenAI for embeddings and chat completions.

How to use Mcp_server_client?

—

Key features of Mcp_server_client

• Streamlit frontend for user interaction
• NestJS API with chat, agent, tool agent, and RAG services
• Integration with multiple MCP servers via SSE
• Retrieval-augmented generation using PGVector
• OpenAI API for embeddings and chat completions
• Storage support: AWS S3, PGVector, PostgreSQL

Use cases of Mcp_server_client

• Building a chat application that leverages multiple MCP servers as tools
• Implementing retrieval-augmented generation with a vector database
• Using an agent-based architecture to orchestrate MCP tool calls

FAQ from Mcp_server_client

What technologies are used in the system?

The system uses Streamlit for the frontend, NestJS for the API, MCP servers as tools, OpenAI for AI services, and storage backends including AWS S3 (via LocalStack), PGVector, and PostgreSQL.

How do MCP servers communicate with the API?

The Tool Agent Service in the NestJS API communicates with MCP servers via SSE (Server-Sent Events) over /sse endpoints.

What storage backends are supported?

The architecture supports AWS S3 (via LocalStack), PGVector (a PostgreSQL vector extension), and standard PostgreSQL.

Does the system use OpenAI?

Yes, the OpenAIService interacts with the OpenAI API for chat completions and embeddings. The RAG service uses embeddings, and the Agent service uses chat completions.