🤖 Building an AI Chat Widget with MCP

Adding an AI-powered assistant to my portfolio I've been exploring the Model Context Protocol (MCP) — Anthropic's open standard for connecting AI models to external data sources. The idea is simple: instead of pasting context into a chat window, you give the AI structured access to your data through tools. I thought it would be a good fit for my portfolio — I have 20 blog posts, several projects, and an about page. Why not let visitors ask an AI assistant about any of it? This post covers how I built two things: an MCP server for Claude Code (local development) and a streaming chat widget for the website (production). 🧩 What is MCP? MCP (Model Context Protocol) is a JSON-RPC-based protocol that lets AI models call "tools" — functions that retrieve or manipulate data. Think of it like giving the AI a set of APIs it can call when it needs information. For example, if someone asks "How do I set up a Jellyfin server?", the AI can: Call a search_blog_posts tool with the query "jellyfin serv