Summary

When a scheduled prompt fires (/every or /after), it interrupts and stops the existing task queue from processing. The queue doesn't resume until the schedule time comes up again, which could be hours or days away.

Steps to Reproduce

1. Start an interactive Copilot CLI session with experimental mode enabled
2. Queue up multiple tasks (e.g., 15 items in the inbox/queue)
3. Schedule a prompt with /every 1h check for new issues
4. Let the queue start processing tasks
5. Wait for the scheduled prompt to fire
6. Observe that after the scheduled prompt completes, the queue does NOT resume processing the next queued item

Expected Behavior

After the scheduled prompt fires and completes, the runtime should automatically resume draining the queue - popping the next item and processing it. The scheduled prompt should not permanently halt the queue.

Actual Behavior

The scheduled prompt interrupts the queue and the queue remains halted until the next scheduled time arrives. This can mean the queue is stuck for hours or days if the schedule interval is long.

Environment

• Copilot CLI version: latest experimental
• Mode: experimental mode with /every scheduling
• OS: Windows 11

Impact

• High: Blocks all queued work for extended periods
• Affects workflows that depend on queue processing (e.g., batch PR reviews, issue triage, test runs)
• Makes /every scheduling unreliable when you also have a queue of tasks

Suggested Fix

Add queue resumption logic after scheduled prompt completion:

function onScheduledPromptComplete(scheduleId) {
  if (queue.hasNext()) {
    dispatchNextQueueItem()
  }
}

Alternative: Add a config option resumeQueueAfterSchedule: true to control this behavior.

Additional Context

The user explicitly requested this behavior: when an AI agent schedules its own prompts, it should not kill in-flight work. The scheduler should either:

• Resume the queue after the scheduled prompt completes, OR
• Warn the user that scheduling will pause the queue and ask for confirmation

This is particularly problematic when the AI autonomously schedules prompts during a session with active queue items.