Rescope Auto Drive for Codex Lab MVP #37

Objective

Make an explicit Auto Drive rescope decision so the MVP does not accidentally recreate the old standalone coordinator surface.

Finish Line

Auto Drive is explicitly retired, deferred, or reduced to Codex-native reusable semantics before any implementation work starts.

Current Status

State: Active. Next action: review restored Auto Drive docs/issues and record which semantics, if any, remain valuable after current goal/review/session surfaces are considered.

Scope

In:

• Classify old Auto Drive behavior as Retire, Defer, or Rewrite semantics later.
• Identify reusable semantics such as long-running goal mode, recovery, review gates, and progress evidence.
• Record explicit non-goals for MVP.

Out:

• No Auto Drive implementation.
• No old coordinator/card/settings port.
• No new background automation loop until Auto Review and app-server compatibility are settled.

Acceptance Criteria

• Decision names old Auto Drive surfaces that should not return in MVP.
• Any retained semantics are mapped to Codex-native surfaces.
• Later implementation, if any, has a separate issue and validation gate.
• Codex Lab MVP dogfood plan #28 sequence no longer treats Auto Drive as a simple late-stage port.

Relationships

Child of #28. Depends on the MVP classification ledger for source evidence.

Validation

Review restored Auto Drive documentation and current Codex goal/review/session behavior; no code validation expected for the decision itself.

Decisions

Auto Drive is a decision issue first, not an implementation slice.

Open Questions

Is any Auto Drive user value still unique after Auto Review proof loop and Desktop/app-server continuity work?