Axenix developer Anton Bogun proposes extending the Doc as Code practice to provide AI agents with deeper project context. While traditional documentation storage in Git ensures version control, it often fails to convey changes in system behavior to AI models.
Modern development involves complex workflows where API methods or request bodies may remain stable while underlying logic shifts. Changes such as new validation steps, event publishing, or asynchronous processing via message brokers often remain invisible to AI agents.
The current approach of granting AI access to OpenAPI specifications, README files, and Markdown pages is insufficient for understanding system dynamics. Developers need to link code, specifications, documentation, and task trackers to ensure AI agents interpret project updates correctly.
