Model Configuration (Console + Agent)

Model setup is split into two responsibilities:

1. Console manages the model pool (provider/model metadata, keys, endpoints).
2. Agent project chooses model.primary in ship.json.

Workflow

1. Configure provider/model in console.
2. Set model.primary in project.
3. Restart agent and validate one real invocation.

Commands

sma console config llm provider list
sma console config llm model list
sma console config set model.primary default
sma agent restart .

Failure cases

• missing model.primary target: agent startup fails.
• invalid provider/key: model invocation fails with auth/network errors.

Related docs

• Model Config Reference