data-model.md should be the ssot when it comes to the application data model (?)

[eumatheusgomes]

My expectation was that once data-model.md was written everything directly or indirectly derived from it would be in sync.

I did several tests using Github Copilot (with Claude Sonnet 4, GPT-5 and Gemini 2.5 Pro) and I always end up with API contracts, API test cases, db migrations and active record models using different field names, which causes Copilot to waste a lot of time debugging and rewriting code.

Example:

• on data-model.md there's an entity with a field named max_daily_quantity
• on api-spec.yaml the related entity response schema references the same field as max_daily_usage
• the related API test case checks for a daily_quantity_limit in the JSON response
• the related db migration creates a table with a daily_usage_limit column
• the related active record model has a dailyLimitReached method which expects a daily_limit column

It doesn't get everything wrong all the time ofc. This is an extreme scenario that happened once so far (using Claude). But I could not get all the 5 things in sync yet though.

I'd appreciate some advice on how and where to enforce consistency here.