II. Context Scaffolding: Treat Context as a Dependency
AI coding agents require rich, structured context to generate quality output. Manage all context—code, documentation, and team standards—with the same rigor as a critical software library.
- Assemble Local Context: Before prompting, use your IDE’s features (e.g., @codebase) to make the AI aware of the relevant files and dependencies within your project.
- Assemble Team Context: Pull in the team’s shared best practices from the team-ai-directives repository (e.g., @team/rules/style-guides/python_pep8_and_docstrings.md).
- Assemble External Context: For external knowledge, use a dedicated search tool to gather patterns and documentation before asking your primary AI to write code.