How segments work ​
Each segment (one part of the day) has the same predictable structure, so you always know where you are and how to catch up.
What every segment page contains ​
| Section | What it gives you |
|---|---|
| At a glance | The start state, end state, and fallback baseline for the part. |
| Learning goals | What you'll be able to do afterward. |
| Why it matters | The motivation, tied to the Dunder Mifflin story. |
| What you will build | The concrete artifact for the part. |
| Knowledge-share outline | The short intro talk. |
| Lab steps | Copy-paste friendly steps, each with expected output. |
| Checkpoint | A quick check that you reached the end state. |
| Fallback baseline | The known-good version to switch to if needed. |
| Troubleshooting | Common problems and fixes. |
| Optional stretch task | Extra depth for advanced attendees. |
| What we will not cover | The scope boundary for the part. |
The rhythm ​
Every part: a short knowledge-share first, then hands-on work, ending at a checkpoint. If your hands-on attempt doesn't finish, switch to the fallback baseline and you'll start the next part in the right place.
The segments ​
- Part 1, Framing & first agent
- Part 2, Context engineering
- Part 3, Memory
- Part 4, Writing useful specs & personas
- Part 5, MCP fundamentals
- Part 6, Build the MCP server
- Part 7, Agent Framework & first MCP-connected agent
- Part 8, Native tools & A2A handoff
- Part 9, Human-in-the-loop & condition checker
Authoring a new segment?
Start from snippets/segment-skeleton.md, which encodes the required structure above.