updates and fixes labs 1 - 4

- The GitLab Swag Shop Flows project exists in your namespace with the pre-built file structure visible in the repository.

- Seven foundational flows are enabled under **Settings > General > GitLab Duo features**.

- The Pages URL opens the live Swag Shop with no featured products on the homepage.

- The agent created an issue from the markdown text.

## Task B: Use a Foundational Flow to Create a Merge Request

   >**Alternative trigger:** You can also trigger the Code Review Flow by typing /assign_reviewer @GitLabDuo in any comment box on the merge request.

### Task C.3: Observe the Flow in Progress

1. Navigate to **Automate > Sessions** in the left sidebar.

1. Locate the new session for the Code Review Flow and open it.

1. Click the **Activity** tab and observe the steps executing in real time. 

   >**What you're seeing:** The flow runs in three steps. First it builds review context, reading the merge request diff, the changed files, and related files like style.css and index.html. Then it reasons over what it found, checking the changes against any custom instructions configured for the project. Finally it generates its recommendations. By the time it posts comments on the merge request, it has read far more than just the changed lines. That broader context is what gives the review its depth.

### Task C.3: Review the Code Review Output

1. Wait for the session status to change to **Finished** before proceeding.

1. From the **Overview** tab, scroll to the **Activity** section. 

### Task C.3: Review the Code Review Output

1. Observe the comments in real time.

1. Return to the merge request and open the **Overview** tab.

   >**What you're seeing:** GitLab Duo Code Review automatically reviews your merge request when requested as a reviewer. It analyzes the changes in the diff and posts a summary comment describing what it found, along with inline suggestions directly on the affected lines of code. The suggestions focus on things like code quality, maintainability, and best practices. In this case, it flagged hardcoded values that should use design tokens and a missing CSS property for proper layering.

1. Read through the comments posted by GitLab Duo. For each one, consider whether you agree with the suggestion and how you would act on it before merging.

### Expected Output: Task C

- GitLab Duo appears in the Reviewers section of the merge request.

- The Code Review Flow session under **Automate > Sessions** has a status of Finished.

- At least one comment from GitLab Duo is visible on the merge request Overview tab.

- At least one comment from GitLab Duo is visible on the merge request **Overview** tab.

## Task D: Review and Merge the Merge Request

1. Navigate to **Plan > Work items** and confirm three issues have been created.

   >**Note:** You will see a total of four issues in your project. One that was created in lab 1, and three created by Duo Agentic chat. If the new issues are not yet visible, manually refresh the page.

   >**Note:** You will see three issues in your project that were created by Duo Agentic chat. If the new issues are not yet visible, manually refresh the page.

1. Navigate to **Code > Merge Requests** and confirm three merge requests have been created, each linked to its corresponding issue.

   Analyze this project's structure and suggest improvements for maintainability and scalability.

   ``` 

1. Review the recommendations returned by Agentic Chat.

1. Review the recommendations returned by agentic chat.

1. Follow up in the same conversation with: 

1. Open one of the flagged issues and confirm you are now listed as the assignee.

1. Return to Agentic Chat and re-run the original prompt: 

1. Return to agentic chat and re-run the original prompt: 

   ```prompt

   Which issues are missing estimates, due dates, or assignees?

Sometimes the right agent for your team doesn't exist yet. Custom agents let you define exactly what an agent knows, what it can do, and who can access it — built for your codebase and your team's specific needs.

Every question since joining has required interrupting someone. There's no good way to ask about the codebase without pulling a teammate away. You decide to fix that, and while you're at it, you start thinking about what it would take to make issue quality automatic rather than something the team has to remember. That's a flow, and it's Lab 4. Right now, you build the tool for the next person who joins.

Every question since joining has required interrupting someone. There's no good way to ask about the codebase without pulling a teammate away. You decide to fix that and build a tool for the next person who joins.

In this lab, you will build the Swag Shop Onboarding Agent: a custom agent that knows the project, answers questions, and creates issues when a new developer finds something worth fixing.

   Helps new developers get up to speed on the DAP Swag Shop codebase, answers questions about the project architecture, and creates issues for improvements or bugs they discover while onboarding.

   ```

   > **Note:** The description appears in the AI Catalog and in the agent dropdown in Agentic Chat. It helps teammates understand what the agent does and when to use it. Keep it concise and specific.

   > **Note:** The description appears in the AI Catalog and in the agent dropdown in agentic chat. It helps teammates understand what the agent does and when to use it. Keep it concise and specific.

1. For **Visibility**, select **Private**.

Our `issue_reviewer` component is an `AgentComponent`. An `AgentComponent` uses a LLM to process inputs and generate responses. Like our previous component, we provide a set of inputs to use. In this case, the `create_issue_note` toolset is used, which will create an issue note. We log tool success, failure, as well as the final answer produced by the component. Since this component uses an LLM, we provide a prompt to use, defined by the `prompt_id` field. The prompt associated with this ID is defined below.

### The Prompt

#### The Prompt

Each prompt is given a prompt ID. You will see our prompt has an ID of `issue_review_prompt`, which matches the prompt ID given to the `issue_reviewer` component. 

1. A comment will appear on the issue confirming that a new flow session has begun, with a link to the session progress.

1. You can also navigate to **Automate > Sessions**, locate the session for the Swag Shop Issue Reviewer, and open it.

  >**Note:** You can also navigate to **Automate > Sessions**, locate the session for the Swag Shop Issue Reviewer, and open it.

1. Click the **Activity** tab and observe the steps executing in real time: issue fetch, data analysis, and comment generation.