What problem does it solve? After routing decides which agent handles a task, coordinators still need to decide how much ceremony that task deserves. This Skill provides the complete decision table for choosing between Direct, Lightweight, Standard, and Full response modes, preventing both over-engineering simple requests and under-serving complex ones. ## Core Features & Use Cases - Four-Tier Decision Table: Maps task complexity to Direct (coordinator answers instantly), Lightweight (minimal single-agent spawn), Standard (full single-agent ceremony), or Full (multi-agent parallel fan-out) modes with target latency ranges. - Exemplar Prompts per Mode: Concrete example requests for each tier, from status checks answered directly to multi-agent "Team" requests touching 3+ domains. - Lightweight Spawn Template: A ready-to-use minimal agent spawn template that skips charter, history, and decisions reads, plus an explore-agent variant for read-only queries. - Upgrade Rules: Explicit bias-toward-upgrading guidance, including the rule to never downgrade mid-task. - Use Case: A user asks "Fix the typo in README" — the coordinator loads this skill, classifies it as Lightweight, and spawns one agent with the minimal template instead of the full ceremony. ## Quick Start Load this skill after routing has identified the agent, then use the decision table to pick the response mode for the user's request.
