Non-interactive CLI UX gap

What feature would you like to see?

1. Return a more specific error when --model does not resolve to a configured local
   model.
   Example:
   • Requested model is not configured locally
   • optionally include the locally configured model names
2. Add a simple command to list configured local models and the current default model.
3. Add a short docs section specifically for unattended / non-interactive usage covering:
   • authentication
   • configured default model
   • explicit model selection
   • validation of local model names before automation use

This would make CLI automation setup more reliable and much easier to diagnose.

Additional information

Non-interactive CLI UX gap:
LLM not set is too vague when --model is not a valid configured local model.

Title: Non-interactive CLI UX gap: LLM not set is too vague when --model is not a
valid configured local model

We encountered this while integrating Kimi Code CLI into unattended automation.

Observed behavior:

• authentication completed successfully
• non-interactive execution with the default configured model worked
• non-interactive execution with an explicit --model value failed with:
  • LLM not set

The current error is too ambiguous. It does not make clear whether the problem is:

• no configured default model
• an explicit --model value that is not defined in the local config
• successful login without a usable configured model entry for non-interactive execution

Why this matters:

• for automation, “logged in” is not sufficient
• operators need a deterministic way to confirm which explicit --model values are
  actually valid in the local configuration
• LLM not set slows down diagnosis because it hides the distinction between auth state

Suggested improvements:

1. Return a more specific error when --model does not resolve to a configured local
   model.
   Example:
   • Requested model is not configured locally
   • optionally include the locally configured model names
2. Add a simple command to list configured local models and the current default model.
3. Add a short docs section specifically for unattended / non-interactive usage covering:
   • authentication
   • configured default model
   • explicit model selection
   • validation of local model names before automation use

This would make CLI automation setup more reliable and much easier to diagnose.

Inspect the CLI handling for --model, local model configuration, and existing non-interactive commands, then review the documentation structure. Confirm the current failure path and configuration format first; done means a distinct invalid-model error, a command showing configured and default models, and documented unattended usage covering authentication and model validation.