[Feature] Built-in batch agent orchestration: queue → dispatch → monitor loop

Problem

Power users frequently run multiple sub-agents in parallel to process a queue of GitHub issues or tasks. The current workflow requires:

1. Manually crafting each sub-agent invocation
2. Switching between tmux windows to check progress
3. Manually detecting failures and re-dispatching
4. No unified view of what all agents are doing

This creates high coordination overhead and limits how effectively users can delegate parallel work.

Proposed Solution

A built-in batch orchestration mode accessible via a command like /batch or through a structured workflow:

/batch <task-list-file-or-issue-query>

Core capabilities needed:

• Queue tasks — accept a list of tasks (GitHub issues, file paths, prompts)
• Dispatch agents — spawn a sub-agent per task automatically
• Monitor status — show a unified status view (task, agent, state, last output)
• Handle failures — detect failed agents and surface them for retry or skip
• Aggregate results — collect outputs/summaries when all agents complete

Use Case

"Here are 12 open GitHub issues. Work through all of them with one agent per issue. Show me progress and flag anything that needs my attention."

Currently achievable only via manual tmux wrangling. A native orchestration layer would make this a first-class workflow.

Related

The agent-inspector skill partially addresses monitoring but requires manual setup and has no dispatch capability.