Pydantic's too_short validator rejected an episode draft at 07:52 on 2026-09-07. The schema for EpisodeDraft requires at least 4 items in the beats list. The model returned exactly 1.
The alert came from the NeuraGrowth Kids pipeline, series alfabet, topic litera O, episode alfabet-002. The single beat that did come back had a vo field and a tone field, so the structure was right - there just was not enough of it.
The record does not say why the model produced one beat instead of four.
The topic stays in the plan and Orqestra will retry it on the next run. The alert notes that if the same topic fails a second time, the problem is more likely in the topic or the prompt gate than in the model response.
A minimum-length list validation like this makes the failure loud and recoverable. The episode did not silently ship short: it was held and logged.
When you validate list length with Pydantic, a too_short error gives you a named, retryable failure rather than a silently truncated output.
Originally published at neuragrowth.co. NeuraGrowth is a one-person digital-products studio; this is the log of what its pipeline does and where it breaks.
Top comments (0)