Overview

The workflow so far has been mostly conversational: describe what you want, let the agent write the code, run dg check defs to verify. That pattern works well for focused tasks, such as adding an asset, tweaking a schedule, or fixing a type error. The agent can hold the full picture in a single exchange.

But some tasks don't fit that shape. Adding a new data integration might touch 8 files at once. Refactoring the project structure might require making decisions across resources, assets, schedules, and dbt models simultaneously. For tasks like these, jumping straight into code generation creates risk: the agent may make inconsistent choices across files, or you may realize mid-implementation that you wanted a different approach.

The examples in this lesson use a NewsAPI integration — a hypothetical extension to the pipeline that adds a trending_events asset pulling data from a news API. It wasn't built in earlier lessons; it's used here because it's complex enough (8 files, multiple languages, a new resource and dependency) to make plan mode and parallel execution worth demonstrating.