github-samples · GeekTrainer · Jul 27, 2026
diff --git a/docs/_images/cli-5-agent-skill.png b/docs/_images/cli-5-agent-skill.png
diff --git a/docs/cli/0-prerequisites.md b/docs/cli/0-prerequisites.md
@@ -49,8 +49,6 @@ The creation of the codespace will take several minutes, although it's still far
 > [!NOTE]
 > This workshop is built to run inside a codespace or local [dev container][dev-containers]. Both ensure the environment has all the necessary prerequisites installed for a smooth experience. If you'd prefer to run it locally, open the cloned repository in VS Code and select **Reopen in Container** when prompted — VS Code will build the same dev container the codespace uses.
 
-[codespaces]: https://github.com/features/codespaces
-[dev-containers]: https://code.visualstudio.com/docs/devcontainers/containers
 ## Summary
 
 Congratulations, you have created a copy of the lab repository! You also began the creation process of your codespace, which you'll use when you begin working with Copilot CLI.
@@ -68,3 +66,5 @@ Let's install Copilot CLI and authenticate it with your GitHub account. Continue
 [template-repository]: https://docs.github.com/repositories/creating-and-managing-repositories/creating-a-template-repository
 [codespaces-quickstart]: https://docs.github.com/codespaces/getting-started/quickstart
 [next-lesson]: ../1-install-copilot-cli/
+[codespaces]: https://github.com/features/codespaces
+[dev-containers]: https://code.visualstudio.com/docs/devcontainers/containers
diff --git a/docs/cli/2-custom-instructions.md b/docs/cli/2-custom-instructions.md
@@ -66,10 +66,6 @@ There isn't one specific way to create instructions files, just as there isn't o
 >
 > Looking for templates or a starting point? Explore [awesome-copilot][awesome-copilot], a repository full of instruction files, custom agents, and other resources.
 
-[ui-instructions]: https://github.com/github-samples/tailspin-toys/blob/main/.github/instructions/ui.instructions.md
-[astro-instructions]: https://github.com/github-samples/tailspin-toys/blob/main/.github/instructions/astro.instructions.md
-[awesome-copilot]: https://github.com/github/awesome-copilot
-[custom-instructions-support]: https://docs.github.com/copilot/reference/custom-instructions-support
 ## Explore the custom instructions files in this project
 
 Take a moment to read the instruction files this repository ships with — there's one core `copilot-instructions.md` and a collection of `*.instructions.md` files for various tasks. Open these in your editor or the GitHub web UI.
@@ -105,34 +101,23 @@ You'll be making code changes, so create a branch to work in.
 
 To see the impact of custom instructions, start by generating code with the current instructions in place. Later, you'll update the file and run a follow-up prompt.
 
-> [!TIP]
-> **Start a Copilot CLI session**
->
-> Before you start the exercises below, return to your codespace and open a terminal (<kbd>Ctrl</kbd>+<kbd>\`</kbd> if one isn't already open). Then start Copilot CLI with `--yolo` and `--enable-all-github-mcp-tools`:
->
-> ```bash
-> copilot --yolo --enable-all-github-mcp-tools
-> ```
->
-> To pick up your most recent session for this project instead of starting fresh, run `copilot --yolo --enable-all-github-mcp-tools --continue`. If Copilot CLI is already running from an earlier exercise, send `/clear` to start a clean conversation.
->
-> `--enable-all-github-mcp-tools` enables the read/write GitHub MCP tools for the current session, so Copilot can read your backlog and open pull requests during the workshop flow.
-
 > [!CAUTION]
 > `--yolo` enables full automatic permissions (`--allow-all-tools`, `--allow-all-paths`, and `--allow-all-urls`). Use it only in an isolated environment like a Codespace or VM, and never alias it as your default for day-to-day development. See [Allowing and denying tool use][allow-all-warning] for details.
 
-[allow-all-warning]: https://docs.github.com/copilot/how-tos/copilot-cli/use-copilot-cli/allowing-tools
-1. Make sure your Copilot CLI session is running from the **repository root** so it picks up `.github/copilot-instructions.md` automatically.
-2. At the Copilot CLI prompt, ask it to generate the publishers helper that the filtering UI will use:
+Running Copilot CLI from the **repository root** ensures it picks up `.github/copilot-instructions.md` automatically. `--enable-all-github-mcp-tools` turns on the read/write GitHub MCP tools so Copilot can read your backlog and open pull requests later in the workshop.
+
+1. Return to your codespace. If you closed it, navigate to your repository on GitHub.com, select **Code** > **Codespaces**, then reopen your existing codespace.
+2. Return to your open Copilot CLI session. If the terminal is closed or you exited Copilot CLI, open a terminal by selecting <kbd>Ctrl</kbd>+<kbd>\`</kbd>, then start it from the repository root by running `copilot --yolo --enable-all-github-mcp-tools`. Trust the project folder if prompted, then run `/models` and select **Auto**.
+3. At the Copilot CLI prompt, ask it to generate the publishers helper that the filtering UI will use:
 
    ```plaintext
    Create a new data-access helper at src/lib/publishers.ts to return a list of all publishers. It should return the name and id for all publishers. Do not run the tests yet.
    ```
 
-3. Copilot CLI will explore the project, propose a plan, and write the file in this `--yolo` session. Monitor the changes in your terminal output, then review in your editor.
-4. Open the generated `src/lib/publishers.ts` in your editor.
-5. Notice the helper is a typed function that takes a `db` client as its first argument and returns a typed array of publishers — that's coming from the data-layer conventions in `.github/instructions/drizzle.instructions.md` (which applies to `src/lib/*.ts`).
-6. Notice the generated code **is missing** TSDoc doc comments and a file-level comment header.
+4. Copilot CLI will explore the project, propose a plan, and write the file in this `--yolo` session. Monitor the changes in your terminal output, then review in your editor.
+5. Open the generated `src/lib/publishers.ts` in your editor.
+6. Notice the helper is a typed function that takes a `db` client as its first argument and returns a typed array of publishers — that's coming from the data-layer conventions in `.github/instructions/drizzle.instructions.md` (which applies to `src/lib/*.ts`).
+7. Notice the generated code **is missing** TSDoc doc comments and a file-level comment header.
 
 > [!CAUTION]
 > Copilot is probabilistic — there's a chance it'll add doc comments even without being told. If that happens, that's fine; the *consistency* improvement after the instruction update is still the takeaway.
@@ -237,3 +222,8 @@ Next, you'll apply these instructions while implementing backlog work in [the ge
 [instruction-files]: https://docs.github.com/copilot/customizing-copilot/about-customizing-github-copilot-chat-responses
 [instructions-best-practices]: https://docs.github.com/enterprise-cloud@latest/copilot/using-github-copilot/coding-agent/best-practices-for-using-copilot-to-work-on-tasks#adding-custom-instructions-to-your-repository
 [copilot-instructions-five-tips]: https://github.blog/ai-and-ml/github-copilot/5-tips-for-writing-better-custom-instructions-for-copilot/
+[allow-all-warning]: https://docs.github.com/copilot/how-tos/copilot-cli/use-copilot-cli/allowing-tools
+[ui-instructions]: https://github.com/github-samples/tailspin-toys/blob/main/.github/instructions/ui.instructions.md
+[astro-instructions]: https://github.com/github-samples/tailspin-toys/blob/main/.github/instructions/astro.instructions.md
+[awesome-copilot]: https://github.com/github/awesome-copilot
+[custom-instructions-support]: https://docs.github.com/copilot/reference/custom-instructions-support
diff --git a/docs/cli/3-generating-code.md b/docs/cli/3-generating-code.md
@@ -24,33 +24,18 @@ One of the best uses of AI is planning. Oftentimes you'll have a good concept of
 
 You'll start the process of creating the new functionality by utilizing plan mode in Copilot CLI.
 
-> [!TIP]
-> **Start a Copilot CLI session**
->
-> Before you start the exercises below, return to your codespace and open a terminal (<kbd>Ctrl</kbd>+<kbd>\`</kbd> if one isn't already open). Then start Copilot CLI with `--yolo` and `--enable-all-github-mcp-tools`:
->
-> ```bash
-> copilot --yolo --enable-all-github-mcp-tools
-> ```
->
-> To pick up your most recent session for this project instead of starting fresh, run `copilot --yolo --enable-all-github-mcp-tools --continue`. If Copilot CLI is already running from an earlier exercise, send `/clear` to start a clean conversation.
->
-> `--enable-all-github-mcp-tools` enables the read/write GitHub MCP tools for the current session, so Copilot can read your backlog and open pull requests during the workshop flow.
-
-> [!CAUTION]
-> `--yolo` enables full automatic permissions (`--allow-all-tools`, `--allow-all-paths`, and `--allow-all-urls`). Use it only in an isolated environment like a Codespace or VM, and never alias it as your default for day-to-day development. See [Allowing and denying tool use][allow-all-warning] for details.
-
-[allow-all-warning]: https://docs.github.com/copilot/how-tos/copilot-cli/use-copilot-cli/allowing-tools
-1. Enter the following prompt into Copilot CLI to create a plan based on the filtering issue:
+1. Return to your codespace. If you closed it, navigate to your repository on GitHub.com, select **Code** > **Codespaces**, then reopen your existing codespace.
+2. Return to your open Copilot CLI session. If the terminal is closed or you exited Copilot CLI, open a terminal by selecting <kbd>Ctrl</kbd>+<kbd>\`</kbd>, then start it from the repository root by running `copilot --yolo --enable-all-github-mcp-tools`. Trust the project folder if prompted, then run `/models` and select **Auto**.
+3. Enter the following prompt into Copilot CLI to create a plan based on the filtering issue:
 
     ```
     /plan Retrieve the issue on the repository related to adding filtering. We already added a publishers helper in src/lib/publishers.ts, so treat that as existing work and plan the remaining updates (games filtering logic, UI, and tests).
     ```
 
-2. Copilot may ask follow-up questions as it builds out its plan. As those arise, answer them based on how you'd build out the functionality.
-3. Once the plan is generated, review the blueprint. You should notice it recommends remaining changes across the data layer and UI, as well as generating tests.
-4. Copilot CLI will offer you the ability to provide additional feedback to the plan. You can cursor down to the indicated section, then type your suggestions. Copilot will incorporate your suggestions into a new version of the plan.
-5. Once you're satisfied, select the option provided by Copilot to begin work building the new feature!
+4. Copilot may ask follow-up questions as it builds out its plan. As those arise, answer them based on how you'd build out the functionality.
+5. Once the plan is generated, review the blueprint. You should notice it recommends remaining changes across the data layer and UI, as well as generating tests.
+6. Copilot CLI will offer you the ability to provide additional feedback to the plan. You can cursor down to the indicated section, then type your suggestions. Copilot will incorporate your suggestions into a new version of the plan.
+7. Once you're satisfied, select the option provided by Copilot to begin work building the new feature!
 
 > [!NOTE]
 > Because Copilot is probabilistic, the exact text and options provided will vary. But you will notice an option to begin building that will read something similar to:
@@ -59,7 +44,7 @@ You'll start the process of creating the new functionality by utilizing plan mod
 >
 > Copilot may offer you the option to enable [autopilot mode](https://docs.github.com/copilot/concepts/agents/copilot-cli/autopilot), as shown in the example above. Autopilot mode allows Copilot CLI to work through a task without waiting for your input after each step. Once you give the initial instruction, Copilot CLI works through each step autonomously until it determines the task is complete. As we are running in a contained environment, we're OK running autopilot and allowing all tools.
 
-6. Copilot will get to work generating the files!
+8. Copilot will get to work generating the files!
 
 > [!NOTE]
 > This operation will likely take several minutes. You will see Copilot edit and create files, update and generate tests, and run all of the tests to ensure everything succeeds. Now's a good time to reflect on what you've explored thus far, or to enjoy a beverage.

diff --git a/docs/cli/4-mcp.md b/docs/cli/4-mcp.md
@@ -36,37 +36,22 @@ There are many other MCP servers available that provide access to different tool
 
 The quickest way to add a server is the interactive `/mcp add` command. You'll register the [Playwright MCP server][playwright-mcp-server], which gives Copilot a browser it can control.
 
-> [!TIP]
-> **Start a Copilot CLI session**
->
-> Before you start the exercises below, return to your codespace and open a terminal (<kbd>Ctrl</kbd>+<kbd>\`</kbd> if one isn't already open). Then start Copilot CLI with `--yolo` and `--enable-all-github-mcp-tools`:
->
-> ```bash
-> copilot --yolo --enable-all-github-mcp-tools
-> ```
->
-> To pick up your most recent session for this project instead of starting fresh, run `copilot --yolo --enable-all-github-mcp-tools --continue`. If Copilot CLI is already running from an earlier exercise, send `/clear` to start a clean conversation.
->
-> `--enable-all-github-mcp-tools` enables the read/write GitHub MCP tools for the current session, so Copilot can read your backlog and open pull requests during the workshop flow.
-
-> [!CAUTION]
-> `--yolo` enables full automatic permissions (`--allow-all-tools`, `--allow-all-paths`, and `--allow-all-urls`). Use it only in an isolated environment like a Codespace or VM, and never alias it as your default for day-to-day development. See [Allowing and denying tool use][allow-all-warning] for details.
-
-[allow-all-warning]: https://docs.github.com/copilot/how-tos/copilot-cli/use-copilot-cli/allowing-tools
-1. In your Copilot CLI session, enter:
+1. Return to your codespace. If you closed it, navigate to your repository on GitHub.com, select **Code** > **Codespaces**, then reopen your existing codespace.
+2. Return to your open Copilot CLI session. If the terminal is closed or you exited Copilot CLI, open a terminal by selecting <kbd>Ctrl</kbd>+<kbd>\`</kbd>, then start it from the repository root by running `copilot --yolo --enable-all-github-mcp-tools`. Trust the project folder if prompted, then run `/models` and select **Auto**.
+3. In your Copilot CLI session, enter:
 
     ```text
     /mcp add
     ```
 
-2. A configuration form appears. Use <kbd>Tab</kbd> to move between fields and fill it in as follows:
+4. A configuration form appears. Use <kbd>Tab</kbd> to move between fields and fill it in as follows:
 
     - **Server Name**: `playwright`
     - **Server Type**: select **Local** (also labelled **STDIO**)
     - **Command**: `npx @playwright/mcp@latest --headless`
     - **Tools**: leave as `*` to allow all of the server's tools
 
-3. Press <kbd>Ctrl</kbd>+<kbd>S</kbd> to save. The server is added and available immediately — no restart required.
+5. Press <kbd>Ctrl</kbd>+<kbd>S</kbd> to save. The server is added and available immediately — no restart required.
 
 The `--headless` flag tells Playwright to run the browser without a visible window, which is required inside a codespace where there's no desktop to display it. Behind the scenes, this writes the server to your `~/.copilot/mcp-config.json` file:
 
@@ -83,13 +68,13 @@ The `--headless` flag tells Playwright to run the browser without a visible wind
 }
 ```
 
-4. Confirm the server is registered and active by listing your MCP servers:
+6. Confirm the server is registered and active by listing your MCP servers:
 
     ```text
     /mcp show
     ```
 
-5. You should see `playwright` listed alongside the built-in `github` server.
+7. You should see `playwright` listed alongside the built-in `github` server.
 
 > [!NOTE]
 > The Tailspin Toys project already uses Playwright for its end-to-end tests, so the browser Playwright needs is typically already installed. If Copilot later reports that a browser is missing, have it run `npx playwright install chromium` and try again.
@@ -133,7 +118,6 @@ Copilot will launch a browser through the Playwright MCP server, walk through ea
 > [!NOTE]
 > The app needs to be running at `http://localhost:4321` for this test. If you stopped the dev server, start it again before sending the prompt. The first time Copilot uses the Playwright MCP server it may need to download a browser — if it reports a missing browser, have it run `npx playwright install chromium` and try again.
 
-[playwright-mcp-server]: https://github.com/microsoft/playwright-mcp
 ## Summary and next steps
 
 Congratulations, you used the Playwright MCP server to manually test your feature with Copilot CLI! To recap, you:
@@ -156,3 +140,4 @@ Now that you've confirmed the feature works, you can continue to the next exerci
 [mcp-blog-post]: https://github.blog/ai-and-ml/llms/what-the-heck-is-mcp-and-why-is-everyone-talking-about-it/
 [github-mcp-server]: https://github.com/github/github-mcp-server
 [cli-add-mcp]: https://docs.github.com/copilot/how-tos/copilot-cli/customize-copilot/add-mcp-servers
+[playwright-mcp-server]: https://github.com/microsoft/playwright-mcp