Skip to content

My workshop notes ​

Keep a running notebook alongside the workshop using GitHub Gist, a lightweight, version-controlled notepad that lives on your GitHub account.

Once saved, clicking My workshop notes in the account menu will open your gist in a new tab.

Set up in 30 seconds ​

  1. Click Copy on the template block below.
  2. Open gist.new in a new tab (you are already logged into GitHub).
  3. Set the filename to workshop-notes.md.
  4. Paste the template, set visibility to Secret, and click Create secret gist.
  5. Copy the Gist URL from your browser and paste it into the box above. The account menu will open your notes directly from then on.

Your Gist auto-saves every edit and keeps a full revision history, so you can always see what you wrote at any point during the day.

TIP

Gist renders Markdown, so checkboxes, tables, and headings all look exactly as they do here. Switch between Edit and the Preview tab in Gist to read your notes cleanly.

Template ​

markdown
# Agentic AI Training workshop notes
> Accelerator Program · Dunder Mifflin package manager scenario

---

## Part 1 · Framing & first agent

- **My agent endpoint:**
- **Key takeaway:**
- **One thing that surprised me:**
- **Questions to follow up:**

---

## Part 2 · Context engineering

- **My Excalidraw context map:** [paste link]
- **One thing I decided to INCLUDE and why:**
- **One thing I decided to EXCLUDE (safety/privacy) and why:**
- **RAG candidates I identified:**
- **Tool calls I identified:**
- **My copilot-instructions.md branch/commit:**
- **Questions to follow up:**

---

## Part 3 · Memory

- **My `order_memory.py` location:**
- **What is stored and why:**
- **Session scope confirmed (new session doesn't see previous session's order):** yes / no
- **Unit tests passing:** yes / no
- **One memory decision that surprised me:**
- **Questions to follow up:**

---

## Part 4 · Specs & personas

- **My domain:**
- **My constitution prompt:** [paste it here]
- **My `/speckit.specify` prompt:** [paste it here]
- **My `/speckit.plan` prompt:** [paste it here]
- **My spec location (`specs/NNN-*/spec.md`):**
- **Statuses I defined (list them):**
- **Roles I defined (list them):**
- **Domain event types I named (at least 6):**
- **One persona and its hard boundary:**
- **One requirement the AI generated that I wouldn't have written myself:**
- **My `plan.md` location:**
- **My `tasks.md` location:**
- **Questions to follow up:**

---

## Part 5 · MCP fundamentals

- **My MCP design (link or file location):**
- **Capabilities I mapped to resources:**
- **Capabilities I mapped to prompts:**
- **Capabilities I mapped to tools (include read-only / approval-required label):**
- **One thing MCP solves that plain tool calling doesn't:**
- **Questions to follow up:**

---

## Part 6 · Build the MCP server

- **My MCP server branch/commit:**
- **Tools I implemented:**
- **Prompt I implemented:**
- **Resource I exposed:**
- **One successful tool call I validated:**
- **One failed tool call (and its error):**
- **Questions to follow up:**

---

## Part 7 · Agent Framework & first MCP-connected agent

- **My agent branch/commit:**
- **MCP capability my agent called:**
- **Next operational step it proposed:**
- **One case where the agent should ask for clarification:**
- **Questions to follow up:**

---

## Part 8 · Native tools & A2A handoff

- **Native tool I wired up:**
- **The A2A handoff contract I designed:**
- **Questions to follow up:**

---

## Part 9 · Human-in-the-loop & condition checker

- **The condition I implemented:**
- **What the human review payload looks like:**
- **Questions to follow up:**

---

## After the day

- **Top three things I want to take back to my team:**
  1.
  2.
  3.
- **One thing I want to build next:**
- **Certifications I am considering:** GH-300 / GH-600
- **Resources I want to read:** see [References](/references/)