2025-09-07
Building a Simple Agent with MCP, Conversation history, Streaming and Smart Frontend using Locally Deployed Mistral over vLLM
Today we'll create an AI Agent with conversation history that can use an MCP Server (we'll write it ourselves), working with a Language Model deployed locally - Mistral over vLLM, all wrapped in FastAPI with Streaming support in Vercel AI SDK v5 format, connected to a frontend that handles collapsible Tool Calls, Streaming and persistence. Essentially, we'll make a ChatGPT clone.
Czytaj dalej