ask-more
Multi-LLM consultation for complex questions. Trigger when user sends "/ask [question]", "/ask" (no args, consult on recent conversation), "/ask --preset NAME [question]", or "/ask --compare [question]" (raw side-by-side, no synthesis). Spawns parallel subagents with different models, collects answers, extracts consensus/differences/conflicts, returns structured diff report with actionable synthesis. Use for second opinions, blind spot detection, and decision support.
