Retone repo into a self-guided walkthrough

.github/ISSUE_TEMPLATE/demo-task.md → .github/ISSUE_TEMPLATE/practice-task.md

@@ -1,8 +1,8 @@
 ---
-name: Demo task
-about: A scoped task for driving a GitHub Copilot app session during the Lunch & Learn
+name: Practice task
+about: A scoped task for driving a GitHub Copilot app session while following the walkthrough
 title: ""
-labels: demo
+labels: practice
 ---
 
 ## Goal

CONTRIBUTING.md

@@ -3,23 +3,23 @@
 Thanks for helping make this a better resource for learning the GitHub Copilot app! 🎉
 
 This repo is a community **learning resource**, not an official GitHub product. The goal is a
-workshop that's accurate, easy to run, and friendly to newcomers.
+self‑guided walkthrough that's accurate, easy to follow, and friendly to newcomers.
 
 ## Ways to contribute
 
 - **Fix or clarify the walkthroughs** in [`docs/`](docs/) — especially when the app's UI or docs
   change.
 - **Improve the sample app or exercises** in [`sample-app/`](sample-app/) and
   [`exercises/`](exercises/).
-- **Add a new exercise** that demonstrates an app feature not yet covered.
+- **Add a new exercise** that exercises an app feature not yet covered.
 - **Report issues** — out‑of‑date steps, broken links, confusing instructions.
 
 ## Ground rules
 
 - Keep the sample app **small, dependency‑light, and beginner‑friendly**. Resist scope creep.
 - Keep `main` **green**: every change should keep `npm test` and `npm run build` passing.
-- The four intentional gaps in the sample app are **load‑bearing** for the demos — don't fix them
-  in `main`; they each have an exercise.
+- The four intentional gaps in the sample app are **load‑bearing** for the exercises — don't fix
+  them in `main`; they each have an exercise.
 - Cross‑check product claims against the
   [official docs](https://docs.github.com/copilot/how-tos/github-copilot-app/getting-started) and
   link to them rather than restating volatile details.

README.md

@@ -1,6 +1,6 @@
 # Learn the GitHub Copilot app
 
-A hands‑on, copy‑paste‑able workshop for walking technical developers through the
+A hands‑on, self‑paced walkthrough for getting real experience with the
 **[GitHub Copilot app](https://github.com/features/ai/github-app)** — the agent‑driven
 desktop app for macOS, Windows, and Linux that
 [went GA on June 17, 2026](https://github.blog/changelog/2026-06-17-github-copilot-app-generally-available/).
@@ -10,41 +10,42 @@ desktop app for macOS, Windows, and Linux that
 > and worktree; review the diff, validate in the integrated terminal and browser, and open a
 > pull request that uses your team's existing checks — all without context‑switching.
 
-This repo gives a facilitator everything needed to run the session and gives attendees a real,
-runnable codebase to drive the agent against.
+Clone this repo, open it in the app, and work through the modules at your own pace. You get a
+real, runnable codebase to drive the agent against and a step‑by‑step path from your first quick
+chat to shipping a reviewed pull request.
 
 ---
 
 ## Who this is for
 
-- **Audience:** technical developers (comfortable with Git, a terminal, and a typical web stack).
-- **Format:** ~45 minutes, demo‑led, with optional hands‑on exercises.
-- **Outcome:** attendees can install the app, run sessions in the three modes, review and ship a
-  PR, run sessions in parallel, and know where the advanced features (canvases, automations,
+- **You:** a developer comfortable with Git, a terminal, and a typical web stack.
+- **Time:** about an hour end to end, or pick individual modules.
+- **What you'll be able to do:** install the app, run sessions in the three modes, review and ship
+  a PR, run sessions in parallel, and know where the advanced features (canvases, automations,
   MCP/BYOK, sandboxes) live.
 
 ## What's in here
 
 | Path | What it is |
 | --- | --- |
-| [`docs/facilitator-guide.md`](docs/facilitator-guide.md) | Minute‑by‑minute run sheet for the 45‑minute session |
-| [`docs/00-prerequisites.md`](docs/00-prerequisites.md) | Install + sign‑in checklist to send attendees beforehand |
+| [`docs/walkthrough.md`](docs/walkthrough.md) | Start here — the suggested path through the modules |
+| [`docs/00-prerequisites.md`](docs/00-prerequisites.md) | Install + sign‑in checklist to do first |
 | [`docs/01-quick-chat.md`](docs/01-quick-chat.md) | Module 1 — Quick chats |
 | [`docs/02-first-session.md`](docs/02-first-session.md) | Module 2 — Your first agent session (modes) |
 | [`docs/03-review-and-pr.md`](docs/03-review-and-pr.md) | Module 3 — Review the diff, verify, open a PR |
 | [`docs/04-parallel-sessions.md`](docs/04-parallel-sessions.md) | Module 4 — Parallel sessions & worktrees |
 | [`docs/05-modes-and-models.md`](docs/05-modes-and-models.md) | Module 5 — Modes, models & reasoning effort |
 | [`docs/06-advanced.md`](docs/06-advanced.md) | Module 6 — Canvases, automations, MCP/BYOK, sandboxes, `/chronicle` |
-| [`docs/cheatsheet.md`](docs/cheatsheet.md) | One‑page reference to project on screen |
-| [`exercises/`](exercises/) | Ready‑made demo tasks + seed issues + solutions |
+| [`docs/cheatsheet.md`](docs/cheatsheet.md) | One‑page reference to keep open while you work |
+| [`exercises/`](exercises/) | Ready‑made tasks + seed issues + solutions |
 | [`sample-app/`](sample-app/) | A tiny Express + TypeScript Task API to drive the agent against |
 
 ## The sample app
 
 [`sample-app/`](sample-app/) is a tiny **Express + TypeScript Task API** with a passing
 [Vitest](https://vitest.dev) + [Supertest](https://github.com/ladjs/supertest) suite. It is
-deliberately small and has a few **intentional gaps** so the agent has real work to do during
-the demo:
+deliberately small and has a few **intentional gaps** so the agent has real work to do as you
+follow along:
 
 1. `POST /tasks` has **no input validation**.
 2. There is **no `DELETE /tasks/:id`** endpoint (and `TaskStore.remove()` is missing).
@@ -61,14 +62,14 @@ npm test      # 8 passing tests
 npm run dev   # http://localhost:3000  (GET /health, /tasks)
 ```
 
-## Run the Lunch & Learn
+## How to follow along
 
-1. **Before the session:** send attendees [`docs/00-prerequisites.md`](docs/00-prerequisites.md)
-   so they arrive with the app installed and signed in.
-2. **Fork or use this repo** so you (and attendees) have issues and PRs to work with. Optionally
-   create the demo issues in one shot — see [`exercises/seed-issues.md`](exercises/seed-issues.md).
-3. **Follow** [`docs/facilitator-guide.md`](docs/facilitator-guide.md) start to finish.
-4. **Project** [`docs/cheatsheet.md`](docs/cheatsheet.md) when you want attendees to follow along.
+1. **Set up first:** work through [`docs/00-prerequisites.md`](docs/00-prerequisites.md) so the app
+   is installed, signed in, and this repo is connected.
+2. **Fork or clone this repo** so you have your own issues and PRs to work with. Optionally
+   create the practice issues in one shot — see [`exercises/seed-issues.md`](exercises/seed-issues.md).
+3. **Work through** [`docs/walkthrough.md`](docs/walkthrough.md) module by module.
+4. **Keep** [`docs/cheatsheet.md`](docs/cheatsheet.md) open as a quick reference while you work.
 
 ## License & contributing
 

docs/00-prerequisites.md

@@ -1,7 +1,6 @@
-# 00 · Prerequisites & setup (send this before the session)
+# 00 · Prerequisites & setup
 
-Send this page to attendees a day ahead so everyone arrives ready. Setup eats demo time —
-front‑load it.
+Get this out of the way first so the rest of the walkthrough goes smoothly. None of it takes long.
 
 ## What you need
 
@@ -47,12 +46,12 @@ The **sidebar** is your map:
 - **Search** — search across your repositories.
 - **Sessions** — active agent sessions grouped by repository, plus **Quick chats**.
 
-## Facilitator pre‑flight
+## Quick checklist
 
 - [ ] Signed in, this repo connected.
-- [ ] Demo issues created (see [`../exercises/seed-issues.md`](../exercises/seed-issues.md)).
+- [ ] Practice issues created (see [`../exercises/seed-issues.md`](../exercises/seed-issues.md)).
 - [ ] `npm install` already run in `sample-app/`.
-- [ ] Decide which **model** you'll demo and confirm it's available to you.
-- [ ] Screen share set so the **right side panel** (where canvases/diffs open) is visible.
+- [ ] Picked a **model** you want to try and confirmed it's available to you.
+- [ ] Noted the **right side panel** — that's where canvases and diffs open.
 
 ➡️ Next: [01 · Quick chat](01-quick-chat.md)

docs/01-quick-chat.md

@@ -1,35 +1,35 @@
 # 01 · Quick chat — explore without a branch
 
-**Time:** ~5 min · **Goal:** show the lowest‑friction entry point. Quick chats let you ask
-questions and brainstorm **without creating a branch or worktree**.
+**Goal:** see the lowest‑friction entry point. Quick chats let you ask questions and brainstorm
+**without creating a branch or worktree**.
 
 ## Why start here
 
 Not every question needs an agent session. Quick chats are perfect for orientation, scoping, and
 "how does this work?" moments — and they cost less than spinning up a full session.
 
-## Do it live
+## Try it
 
 1. In the sidebar, click **+** next to **Quick chats**.
 2. Ask something grounded in this repo:
 
    > Give me an overview of the `sample-app` Task API: its routes, where state lives, and any gaps
    > or TODOs you can find.
 
-3. Follow up to show it's a real conversation:
+3. Follow up to see it's a real conversation:
 
    > Which endpoint is missing input validation, and what would you add?
 
-## Talking points
+## What's happening
 
 - **No side effects.** Nothing is branched or written — it's a conversation.
 - **Repo‑aware.** It can read the connected repo to ground its answers.
 - **Scope before you build.** Use a quick chat to clarify requirements, then open a session to do
   the work. This is the recommended way to keep AI usage efficient.
 - **History is saved.** Chats are listed by name so you can come back to them.
 
-## Land the transition
+## Before you move on
 
-> "Great — Copilot understands the gap. Now let's give it a branch and have it actually fix one."
+Copilot now understands the gap. Next, give it a branch and have it actually fix one.
 
 ➡️ Next: [02 · Your first session](02-first-session.md)

docs/02-first-session.md

@@ -1,9 +1,9 @@
 # 02 · Your first agent session — the centerpiece
 
-**Time:** ~12 min · **Goal:** start a session from an issue, use **Plan** mode to agree on an
-approach, then **Interactive** mode to steer the agent to a real, tested change.
+**Goal:** start a session from an issue, use **Plan** mode to agree on an approach, then
+**Interactive** mode to steer the agent to a real, tested change.
 
-This is the most important module. Slow down and narrate what you're doing and why.
+This is the most important module. Take your time and pay attention to what you're doing and why.
 
 ## The change we'll make
 
@@ -16,8 +16,8 @@ behind it.
 1. In the sidebar, open **My work** and find the issue **"Add input validation to POST /tasks"**
    (create it first with [`seed-issues.md`](../exercises/seed-issues.md) if needed).
 2. Click the issue, then **New session**. The issue context loads automatically.
-3. Below the prompt, choose where the session runs: a **new working tree** (recommended for the
-   demo — it keeps your main checkout clean), your local repo, or a **cloud sandbox**.
+3. Below the prompt, choose where the session runs: a **new working tree** (recommended — it keeps
+   your main checkout clean), your local repo, or a **cloud sandbox**.
 4. Pick a **session mode**, **model**, and **reasoning effort** (more on these in
    [05](05-modes-and-models.md)).
 
@@ -32,11 +32,11 @@ Select **Plan** mode and prompt:
 > `title` should return `400` with a clear JSON error and must not create a task. Add tests
 > covering the happy path and the invalid cases.
 
-The agent proposes a plan. **Read it aloud.** Approve it, or steer it:
+The agent proposes a plan. **Read it carefully.** Approve it, or steer it:
 
 > Keep it dependency‑free — validate in the route handler, don't add a validation library.
 
-**Point to make:** Plan mode front‑loads the disagreements. You catch a wrong approach before any
+**Why it matters:** Plan mode front‑loads the disagreements. You catch a wrong approach before any
 code is written.
 
 ## Step 2 — Interactive mode (steer the work)
@@ -47,7 +47,7 @@ Switch to **Interactive** mode and let it implement. As it works:
 - Nudge it if needed:
   > Also return `400` when `title` is only whitespace, and add a test for that.
 
-**Point to make:** you're collaborating turn by turn — the agent suggests, you approve or redirect.
+**Why it matters:** you're collaborating turn by turn — the agent suggests, you approve or redirect.
 
 ## Step 3 — Confidence check with rubber duck (optional)
 

docs/03-review-and-pr.md

@@ -1,7 +1,7 @@
 # 03 · Review the diff, verify, and open a PR
 
-**Time:** ~7 min · **Goal:** treat agent output like any teammate's — read the diff, verify it
-runs, then ship it through your normal GitHub flow.
+**Goal:** treat agent output like any teammate's — read the diff, verify it runs, then ship it
+through your normal GitHub flow.
 
 ## Step 1 — Read the diff
 
@@ -11,7 +11,7 @@ Open the session's **diff** view and walk the changes:
 - Is the validation logic correct and the error shape sensible?
 - Are the new tests meaningful, not just present?
 
-**Point to make:** the diff is the contract. You review the change, not the chat transcript.
+**Why it matters:** the diff is the contract. You review the change, not the chat transcript.
 
 ## Step 2 — Verify in the integrated terminal
 
@@ -29,7 +29,7 @@ npm run dev
 # in another terminal tab:
 curl -s -X POST http://localhost:3000/tasks -H "content-type: application/json" -d "{}"
 # expect HTTP 400 with a JSON error
-curl -s -X POST http://localhost:3000/tasks -H "content-type: application/json" -d "{\"title\":\"Demo\"}"
+curl -s -X POST http://localhost:3000/tasks -H "content-type: application/json" -d "{\"title\":\"Write docs\"}"
 # expect HTTP 201 with the created task
 ```
 
@@ -40,7 +40,7 @@ Open the integrated **browser** canvas and hit:
 - `http://localhost:3000/health` → `{ "status": "ok" }`
 - `http://localhost:3000/tasks` → the tasks you created
 
-**Point to make:** terminal + browser are *inside* the app, so reviewing and validating doesn't
+**Why it matters:** terminal + browser are *inside* the app, so reviewing and validating doesn't
 mean alt‑tabbing across three windows.
 
 ## Step 4 — Open the pull request
@@ -50,9 +50,9 @@ From the session, **create a pull request**. Then:
 - Show the PR uses your repo's **existing checks and merge requirements** — the CI in
   [`.github/workflows/ci.yml`](../.github/workflows/ci.yml) runs on the same change.
 - Show that you can **review the PR and view CI status** from **My work** without leaving the app.
-- Merge it (or leave it open for Q&A).
+- Merge it, or leave it open to revisit later.
 
-## The loop to name out loud
+## The loop, end to end
 
 > *issue → session (mode + model) → review the diff → verify → open PR → check CI → merge.*
 

docs/04-parallel-sessions.md

@@ -1,15 +1,15 @@
 # 04 · Parallel sessions & worktrees
 
-**Time:** ~7 min · **Goal:** show the app's signature move — multiple agent sessions running at
-once, each isolated on its own branch and git worktree.
+**Goal:** see the app's signature move — multiple agent sessions running at once, each isolated on
+its own branch and git worktree.
 
 ## Why this matters
 
 Most "AI in the editor" tools are single‑threaded: one conversation, one working copy. The Copilot
 app runs **many sessions in parallel**, each in a **dedicated worktree and branch**, so progress on
 one task never clobbers another. You shift from *doing the work* to *directing several streams of it.*
 
-## Do it live
+## Try it
 
 While the PR from Module 3 is building, start two more sessions from issues:
 
@@ -18,22 +18,22 @@ While the PR from Module 3 is building, start two more sessions from issues:
 3. Session B — **"Fix: completing a task doesn't update updatedAt"** ([exercise 3](../exercises/README.md)).
 4. Give each a one‑line prompt and let them run.
 
-Now switch between **all three** sessions in the sidebar (grouped by repository). Call out:
+Now switch between **all three** sessions in the sidebar (grouped by repository). Notice:
 
 - Each session has its **own branch and worktree** — open two diffs side by side to prove the
   isolation.
 - You can give each a **different mode and model** (e.g. Autopilot + a fast model for the small
   DELETE endpoint; Interactive for the subtler bug).
 - Nothing blocks: Session A doesn't wait for Session B.
 
-## A good narration
+## The mental model
 
-> "I'm not waiting on any one agent. The DELETE endpoint is well‑scoped, so I'll let it run on
-> Autopilot. The `updatedAt` bug is subtler, so I'll keep that one Interactive and steer it. Both
-> live in their own worktree, so neither can step on the other — or on the validation PR that's
-> already in review."
+Think of it this way: you're not waiting on any one agent. The DELETE endpoint is well‑scoped, so
+you can let it run on Autopilot. The `updatedAt` bug is subtler, so keep that one Interactive and
+steer it. Both live in their own worktree, so neither can step on the other — or on the validation
+PR that's already in review.
 
-## Cloud sessions (mention)
+## Cloud sessions (good to know)
 
 Sessions can also run in **cloud sandboxes** (public preview) — fully isolated Linux environments
 hosted by GitHub. Great for offloading compute or picking work up from another device. More in

docs/05-modes-and-models.md

@@ -1,7 +1,6 @@
 # 05 · Modes, models & reasoning effort
 
-**Time:** ~2 min on screen · **Goal:** give attendees a mental model for the knobs they'll reach
-for in every session. This is a recap slide more than a demo.
+**Goal:** build a mental model for the knobs you'll reach for in every session.
 
 ## Session modes — how much autonomy
 

docs/06-advanced.md

@@ -1,8 +1,7 @@
 # 06 · Advanced features — a guided tour
 
-**Time:** ~3 min · **Goal:** show attendees the ceiling. Don't deep‑dive; point at each feature,
-say what it's for, and move on. Pick **one** to demo live if time allows (canvases is the crowd
-pleaser).
+**Goal:** see the ceiling. Don't try to master each feature now — learn what each one is for, then
+try **one** that interests you (canvases is a good first pick).
 
 ## Canvases — shared, interactive surfaces
 
@@ -67,4 +66,4 @@ concrete feedback, running on a **different model** than your main session. Invo
 - **Agent skills** (`/`‑commands) to package repeatable workflows.
 - **Voice dictation** to speak prompts instead of typing them.
 
-➡️ Back to the [facilitator guide](facilitator-guide.md) · or the [cheat sheet](cheatsheet.md)
+➡️ Back to the [walkthrough](walkthrough.md) · or the [cheat sheet](cheatsheet.md)

docs/cheatsheet.md

@@ -1,6 +1,6 @@
 # Cheat sheet — GitHub Copilot app
 
-Project this during the session. One screen, the essentials.
+Keep this open while you work. One screen, the essentials.
 
 ## The core loop