What problem does it solve? When an AI architect agent needs expert input on unfamiliar domains, ad-hoc questioning leads to unbounded calls, inconsistent answers, and lost context. This Skill enforces a disciplined consultation protocol that bounds evidence gathering and makes uncertainty explicit. ## Core Features & Use Cases - Bounded SME Consultation: Identifies 1-3 relevant domains from task requirements and calls the swarm's SME agent once per domain, serially, with a hard cap of 3 calls per project phase. - Guidance Caching: Checks .swarm/context.md for cached guidance before consulting and stores new guidance there to avoid redundant calls. - Re-consultation Triggers: Re-consults only when a new domain emerges or significant changes require fresh evaluation. - Use Case: An architect agent planning a feature that touches authentication and database schema consults the SME agent once per domain, caches the guidance, and proceeds with clearly stated uncertainty. ## Quick Start Activate MODE: CONSULT to answer this advisory question by consulting the swarm SME agents for the relevant domains and caching the guidance in .swarm/context.md.
