Working with Code Assistants: The Skeleton Architecture

"Producing quality code and controlling tech debt with AI Code assistants requires structure and guardrails. It is not enough to rely on prompts. Reducing the size context window that the assistant needs to generate code is critical. Vertical Slice architectures reduce context but miss important cross cutting concerns and common approaches across slices. Using the Dependency Inversion Pattern together with Vertical Slices provides the assistant with a template for implementation that constrains the context size and guides the structure of the generated code."

"GitHub CEO Thomas Dohmke recently issued a stark warning: "Either you embrace AI, or get out of this career". But embracing AI doesn't just mean using autocomplete. It means shifting our primary skill from syntax to Systems Thinking- learning to "decompose problems until they are small enough" for the AI to solve. In short: we are all architects now. I have been building an IoT application with device firmware, a back end, and a web UI."

AI code assistants can increase developer velocity but require architecture, structure, and guardrails to avoid uncontrolled changes and technical debt. Reducing the assistant's context window is critical for reliable code generation. Vertical Slice architectures shrink context per slice but omit cross-cutting concerns and shared conventions. Combining Vertical Slices with the Dependency Inversion Pattern provides templates that constrain context size and guide generated structure. Base classes implement nonfunctional requirements and application workflows, forming a Skeleton layer that enforces security, performance, observability, and standardized message publishing. The skeleton can add controls such as message-schema compliance to further constrain and validate model-generated code.

Read at InfoQ