semantic.ai_model_literal: A ModelId is written out anywhere but a setting's default:

A ModelId is written out anywhere but a setting's default:.

Area 21, AI; not declinable; the next step is usually the agent's.

A model is a ModelId setting, so moving to a newer one is a configuration change reviewed like any other; written in code, it could change only with a new build. A setting's default: and test:, and a test's config:, give a setting its value, so they may write one.

example

/// The model drafts fall back on when the usual one is busy.function fallback_model() -> ModelId {    return "openai/gpt-4.1-mini"}

jadpo checkcheck failed: 1 error, 1 file

error "openai/gpt-4.1-mini" writes a model in code

--> main.jadpo:3:12

  |3 |     return "openai/gpt-4.1-mini"  |            ^^^^^^^^^^^^^^^^^^^^^ a `ModelId` written out

reason: A model is a setting, so moving to a newer one is a configuration change reviewed like any other: a ModelId setting gives it a default:, ai: { models: } names it, and a call asks it with model:.
next: Declare a ModelId setting with this default: in config.jadpo, name it in ai: { models: }, and ask it with model: (agent)
rule: semantic.ai_model_literal
help: https://jadpo.dev/docs/errors/semantic.ai_model_literal