lore onboard and writ migrate send the user to lore config model, which does not exist #1015

With no AI provider configured, lore onboard refuses, which is right (the owner, 2026-10-02: "lore onboard should fail until an ai provider"). Its message sends the user to a command that does not exist: "AI provider required; configure with 'lore config model'" (cmd/lore/lore/commands.go:776). The config set on every program is edit, get, list, path, schema, set, unset and validate (docs/architecture/configuration.md § The command surface); nothing registers config model.

The same command is named at cmd/writ/writ/migrate_cmd.go:128 and cmd/writ/writ/migrate/plan.go:206 ("model provider required for migration analysis"), and in cmd/internal/e2e/harness.go:13, :20 and :59.

The owner asked that prompting for a provider be considered: "we should consider prompting for one." The machinery exists. model.EnsureProvider (cmd/internal/model/config.go:170) ends its resolution with an interactive prompt when it is allowed one; lore onboard does not use it, reading lore.model.* through viper instead (#1010).

Found 2026-10-02 at 01634491, checking that every program holds its own defaults (#1009).

Resolution

The refusal stays. Every message names what configures a provider: model.provider in the file, DEVLORE_MODEL_PROVIDER, or --model-provider, which #1010 makes reach every program. Whether lore onboard prompts for a provider, through model.EnsureProvider's interactive step, is decided in this lane's plan.

Plan and design documents

Epic: #441 · Feature: #456