LAB: Building Better Context With Instructions

Every coding tool has some way of adding context to the chat sessions. With Copilot, you can use a directory called .github/copilot-instructions.md. With Claude it's a CLAUDE.md file in the project root, and with Gemini CLI it's GEMINI.md.

Instructions are mostly done with markdown text, however, you do not need to limit yourself to this! AI tools are very good with well-known structures, like JSON, YAML, or even SQL.

In this session, we're going to:

• Improve our Git commit message by adding instructions to both Copilot and Gemini CLI.
• Create a simple project specification and see how Copilot will translate that into code that we can use later in the session.
• Add style guides for our code, databases, and tests to cut down on the size of our prompts and to get back better responses.
• Use JSON and SQL files to help our AI tool create a data access layer for our project, using specific templates that our boss will approve of.