Deep Research로 작성한 "마스터클래스" 마크다운(예: advertising_masterclass.md)을 Claude Code 프로젝트에 두십시오.
모든 에이전트가 모델의 디폴트 대신 실제 도메인 지식을 참조하게 되어, LLM이 아니라 그 문서가 주제 전문가가 됩니다.
Drop a Deep-Researched "masterclass" markdown (e.g. advertising_masterclass.md) into the Claude Code project so every agent references real domain expertise instead of LLM defaults — the doc, not the LLM, becomes the subject-matter expert.
Deep Research로 작성한 "마스터클래스" 마크다운(예: advertising_masterclass.md)을 Claude Code 프로젝트에 두십시오.
모든 에이전트가 모델의 디폴트 대신 실제 도메인 지식을 참조하게 되어, LLM이 아니라 그 문서가 주제 전문가가 됩니다.
에이전트 품질은 큐레이팅된 참조 문서로 확장됩니다 — 단순히 프롬프트를 잘 짜는 일이 아닙니다. Alex의 brain-as-SME 패턴과 직결됩니다: memory/ 트리 자체가 큐레이팅된 SME 문서 세트이고, thought가 그 원자입니다.