GitHub Copilot Chat is an in-IDE conversational assistant that combines dialog-based prompting with Copilot-style coding help. Use it to ask contextual development questions without leaving the coding environment, then refine requirements through follow-up prompts. section 2.4 section 5.4
Treat responses as a capable starting point rather than an authority. Generated SQL, code, and recommendations need review against the intended schema, permissions, constraints, quality standards, and domain requirements before use. section 5.1 section 5.4
A productive workflow is to:
- State the concrete task and relevant constraints, such as a database name, schema, columns, keys, valid values, or required row count. section 5.1
- Inspect the proposed command or code for omissions and assumptions, especially access permissions and relationships between tables. section 5.1
- Run and test the result in the target environment, then ask a focused follow-up question when a design or implementation detail remains unresolved. section 5.1 section 5.4
- Use the conversation to explore alternatives when an initial generation is poorly suited to the task. section 5.1
Building an asset-data setup
For an IT asset-management database, prompts can request a PostgreSQL container, an itam schema, role permissions, tables with primary and foreign keys, sequences, and a constrained synthetic dataset. Review each generated statement: a plausible dataset query or table definition may still require changes to match the application model. section 5.1 section 5.4
