AI Usage

This document outlines where AI is used, where it is explicitly avoided, and how its usage varies across the AniBridge repositories.

For each repository group, this document provides a brief description of its purpose, a summary of where AI is and isn't used, and a rough estimate of the percentage of AI-generated code (excluding tests).

`anibridge/anibridge`

The main repository containing the backend, frontend, core application logic, and deployment configuration.

AI was used heavily during the v2 migration and refactor. Since then, usage has decreased significantly as the codebase has stabilized. It is most prevalent in the frontend and least prevalent in core synchronization logic and provider integrations.

AI is used for	AI is NOT used for
Test generation	Core sync logic
Debugging and optimization	Provider integrations
Implementation review	Database design
Frontend integrations and styling	API design
Web services	Web layout
	Containerization / deployment

AI-generated code: ~40%

`anibridge/anibridge-utils`

Shared helpers and utilities used across the project. Because these are foundational to how everything else behaves, all core implementations are written and maintained by hand. AI is primarily used for reviewing and refactoring existing code, as well as generating tests.

AI is used for	AI is NOT used for
Test generation	Core implementations
Implementation review
Refactor suggestions

AI-generated code: ~20%

`anibridge/anibridge-library-base`

`anibridge/anibridge-list-base`

These repositories define the base classes and interfaces that all library and list providers implement. All hand-coded.

AI-generated code: ~0%

`anibridge/anibridge-plex-provider`

`anibridge/anibridge-jellyfin-provider`

`anibridge/anibridge-emby-provider`

`anibridge/anibridge-anilist-provider`

`anibridge/anibridge-mal-provider`

`anibridge/anibridge-trakt-provider`

Each provider follows the interfaces defined in the anibridge-library-base / anibridge-list-base repositories. Becuase a lot of the code is shared, AI is used to adapt logic between providers.

AI is used for	AI is NOT used for
Test generation	Core provider logic
Debugging	API interactions
Feature adaptation between providers	Data modeling
	Feature implementation

AI-generated code: ~30%

`anibridge/anibridge-mappings`

`anibridge/anibridge-metadata`

ETL pipelines that process and normalize data from several external APIs. This is the most AI-assisted part of the project.

AI is used for	AI is NOT used for
Test generation	Validation logic
Data normalization logic	Schema design
Architectural decisions	API design
ETL orchestration

AI-generated code: ~80%

`anibridge/anibridge-docs`

Documentation for the AniBridge project. All docs are done by hand.

AI-generated code: ~0%