📕 Build AI Agents using LangGraph.js is now out!

Paolo Ricciuti – Svelte ambasador & maintainer of TMPC TypeScript MCP & Senior Software Developer

Today’s episode features Paolo Ricciuti, a Svelte maintainer, Senior Software Developer @Mainmatter. We chat about Model-Context Protocols (MCP), why giving LLMs tools and reliable context matters, and how Paolo built TMCP — a lightweight, composable TypeScript library for building MCP servers.

Paolo walks us through his origin story (games → PHP → Java → photography → web), his day-to-day work at Mainmatter. We dig into TMCP’s design goals (small dependencies, composability, consistent API), the clever “autofixer” approach used in the Svelte/MCP server, and practical tips for using AI to speed up your tasks without losing control.

🔗 Mentioned links

Below are resources and projects we mentioned during the conversation:

🎙 Chapters and topics

  • 0:00 - Intro & welcome
  • 0:53 - Paolo’s origin story: games, PHP, Java, photography
  • 3:06 - Returning to web dev and discovering Svelte
  • 6:59 - Day-to-day: working at Main Matter
  • 10:05 - What is MCP (Model-Context Protocol)?
  • 14:00 - Why Paolo built TMCP (TypeScript MCP library)
  • 15:03 - Problems with the official SDK & design decisions
  • 22:08 - Building TMCP from a minimal JSON-RPC server
  • 25:41 - TMCP design goals: composability, minimal deps, consistent API
  • 31:40 - Using AI to speed up the last 30% (tests, docs, reviews)
  • 39:16 - Why MCP brings back control to developers
  • 40:01 - The Svelte MCP server & the “autofixer” technique
  • 45:00 - Practical results: improving generated Svelte code
  • 52:02 - What makes a great software engineer (curiosity)
  • 56:00 - How Paolo learns and builds things
  • 57:27 - Where to follow Paulo (BlueSky / GitHub) and closing

Enjoy the conversation! If you liked the video, please like, subscribe, and check the pinned comment for direct links to TMCP and the Svelte MCP server.

📖 Build a full trivia game app with LangChain

Learn by doing with this FREE ebook! This 35-page guide walks you through every step of building your first fully functional AI-powered app using JavaScript and LangChain.js

📖 Build a full trivia game app with LangChain

Learn by doing with this FREE ebook! This 35-page guide walks you through every step of building your first fully functional AI-powered app using JavaScript and LangChain.js


Leave a Reply

Your email address will not be published. Required fields are marked *