Machine-readable public data keeps the door open for future assistants while avoiding early write surfaces.

Once the signal shape feels durable, a private API can reuse the same normalization and validation rules instead of inventing a second model.