Problem

Users are tied to the terminal even when Claude's question is not critical. In the vast majority of cases, Claude explicitly lists action options, indicates which one is recommended, and explains why.

Current flow:

Claude: "Do you want option A (recommended) or option B?"
         ↓
[User walks away from computer]
         ↓
[Session stalls indefinitely]

Proposal

Allow Claude to auto-continue the session after a configurable timeout by selecting its own recommended answer.

Proposed flow:

Claude: "Do you want option A (recommended) or option B?"
         ↓
[Configurable timeout, e.g. 5 minutes]
         ↓
Claude: "No response received. Proceeding with option A (recommended)."
         ↓
[Session continues]

Configuration options (suggestions)

// settings.json or similar
{
  "autoContineue": {
    "enabled": true,
    "timeoutMinutes": 5,
    "tokenBudget": 1000,  // optional: budget for deliberation before auto-selecting
    "notifyOnAutoContinue": true  // optional: sound/notification when auto-continued
  }
}

Token budget (optional): For non-trivial decisions, Claude could use subagents or additional reasoning within a token budget to make a better choice. But even a simple "pick my recommendation" would provide huge QoL value.

Use cases

1. Long-running tasks: User starts a multi-phase implementation, steps away for coffee
2. Background work: User wants Claude to proceed with sensible defaults while they focus elsewhere
3. Overnight sessions: Start a large refactoring, let it run overnight with auto-decisions

Value

• Reduces user friction in long sessions
• Leverages Claude's existing recommendation system
• Makes Claude Code more autonomous without sacrificing user control (opt-in feature)

This feature request was co-authored with Claude during a coding session where this exact scenario occurred.