Hi everyone. I'm still pretty new to using GitHub Copilot in my data engineering work, and I'm honestly a bit nervous about it. I work with Airflow DAGs and dbt models, and the idea of an AI suggesting code that might accidentally break a production pipeline is... stressful.
I keep hearing that "context" is everything for getting good Copilot suggestions. Could someone explain, in simple terms, how Copilot decides what context to use? When I'm editing a Python script in VS Code, what files is it actually "reading" to make its suggestions?
For example, if I have a project structure like this:
```
my_project/
├── dags/
│ └── my_dag.py
├── includes/
│ └── sql_queries.py
└── utils/
└── bigquery_client.py
```
When I'm typing in `my_dag.py`, does it look at the open files in my editor? Does it scan the whole project directory? I'm especially worried about it pulling in an old, deprecated utility file and suggesting code from that.
A concrete scenario: I'm writing a task to load data into BigQuery. Will Copilot's suggestions be influenced by the `bigquery_client.py` module I imported, or by similar patterns it sees in other open DAG files? Knowing the boundaries would help me feel more in control and avoid unsafe suggestions.