Set Up GitHub Copilot for Day to Day Pipeline and dbt Code

Copilot ships as an extension, not a separate app, so it lives inside the editor you already use.

1. Open VS Code (or JetBrains, or Visual Studio).
2. Go to the Extensions panel and search for "GitHub Copilot."
3. Install the extension published by GitHub.
4. When prompted, sign in with your GitHub account and authorize the editor.

What you should see: A Copilot icon appears in the status bar or sidebar once signed in, and you're placed on the free plan by default unless your organization already assigned you a paid seat.

Troubleshooting: If your company issues laptops through a managed device policy, the extension marketplace may be restricted. Ask IT or your platform team whether Copilot is already approved and pushed through their tooling instead of installing it yourself.

1. Click the Copilot icon in your editor's status bar.
2. Open the account or settings menu to check your current plan and usage.

What you should see: Free shows a running count toward your monthly completions and premium request limit. A paid seat shows no visible counter, or a much higher one, depending on plan.

Open a dbt model or pipeline script you're already working on.

1. Start typing a new function, model, or SQL block as you normally would.
2. Pause after a line or two. Copilot shows a greyed out suggestion for what typically comes next.
3. Press Tab to accept, or keep typing to ignore it.

What you should see: Suggestions that lean on patterns already present in your open files and repo, not generic boilerplate. A new incremental model, for instance, often gets an is_incremental() block and merge key suggestion pulled from similar models elsewhere in your project.

Troubleshooting: If suggestions look generic or wrong, Copilot may not have enough open context. Open a couple of related files in the same tab group first so it can see the patterns you're following.

Inline suggestions handle "what comes next." The chat panel handles "explain this" or "why is this failing."

1. Open the Copilot chat panel from the sidebar icon or your editor's command palette.
2. With a file open, ask a question about it directly, for example "why would this dbt model fail on an incremental run but pass on a full refresh."
3. Read the response, which references the specific code in your open file.

What you should see: A chat response that quotes lines from your actual file rather than a generic explanation.

Before you lean on Copilot daily, know what it reads.

1. Copilot suggests based on the code open in your editor and, depending on your organization's settings, your indexed repository.
2. Never paste production database credentials, API keys, or customer records into a chat prompt, even to ask for help debugging them. Reference the variable name or a placeholder instead.
3. If your repo has a secrets file, a .env, or credential configs, make sure your editor or repo settings exclude those from whatever Copilot indexes. Check with your platform or security team for how your org has this configured, since exact settings vary by plan and by whether you're on an individual or organization seat.

Why this matters: A pipeline repo often has connection strings, warehouse hostnames, and sometimes stray sample data sitting in test fixtures. None of that belongs in a prompt to a hosted AI model, even a good one.