Dispatch the Research Team on any question. Amplifies a terse prompt into a full multi-specialist investigation with adversarial gates (skeptic + adversary + moderator + evaluator). Use this skill IMMEDIATELY when the user asks to research something, investigate a topic, check what's happening with a technology, compare options, or asks "what's the SOTA on X". Examples: "research vLLM MoE routing", "check HN about agent memory", "what's the best approach for X", "investigate why Y is slow", "compare A vs B vs C".
You are invoking the Research Team (v2.1 protocol, 17 specialists).
Read ~/.claude/agent-memory/research-lead/MEMORY.md (first 200 lines) — these lessons are binding on dispatch decisions.
Read ~/.claude/agents/research/research-lead.md — adopt this as your behavioral contract for this session.
Create the session workspace at <cwd>/.claude/teams/research/<slug>/ where <slug> is derived from the user's question (e.g. "vllm-moe-routing-2026q2").
Execute the full v2.1 protocol:
Run audit gates:
python3 ~/.claude/scripts/audit_evidence.py <slug> --gate=mid-flightpython3 ~/.claude/scripts/audit_evidence.py <slug> --gate=synthesis --strictDeliver the synthesis to the user.