26/06/2026
🤖 Working with LLMs or building AI agents for your dev environment? The terminology is
evolving rapidly, and it can easily feel like guesswork.
To help make sense of the jargon, I've broken down Matt Pocock’s popular "Dictionary of AI
Coding" glossary, detailing the core concepts you need to know.
We look at:
✅ The Model vs. The Harness: Understanding what actually executes your code.
✅ The "Smart Zone" vs. "Dumb Zone": Why your agent starts getting sloppier after long
sessions (and how to fix it).
✅ Tools & MCP (Model Context Protocol): standardizing how agents talk to your files.
✅ DX vs. AX (Agent Experience): How to configure a codebase so AI can read it efficiently.
I've also linked these concepts back to notes in my personal digital garden (wiki.kamath.
cloud).
Read the complete guide here:
Understand the dictionary of AI coding with this guide. Learn key terms like model, harness, context window, and agentic loops.