AI Field Guide

AI Field Guide

The words everyone keeps using (agent, harness, context, MCP), explained with demos you can play with. Switch between Plain English and Developer mode at the top.

How the pieces fit together

Click any piece to jump to its explanation.
Harness (e.g. Claude Code) You Context window System prompt Skills Memory Model(the LLM) Agent loop ↻ Tools Subagents Permissions MCP GitHub Database Slack Vector DB (RAG)
06

Good habits, bad habits

AI is most dangerous when it sounds sure of itself. These are the common ways people get burned, and what to do instead.

07

Check yourself

Eight quick questions. Pick an answer to see why.