Skip to content

Engine-Based Architecture (v6.0)

Newsfork v6.0 — Engine-Based Architecture

Section titled “Newsfork v6.0 — Engine-Based Architecture”

Version: 6.0
Status: Target architecture (planning / phased rollout)


“각 기능은 독립된 엔진으로 분리되어, 일관된 네이밍과 인터페이스를 통해 개발된다”

  • Each engine is developed, deployed, and tested independently.
  • Inter-engine communication is only via Queue or API (no direct shared state).
  • All engines follow the same naming convention and folder structure (nf-{engine}-{resource}-{qualifier}).
  • External exposure is through Zuplo API Gateway.

CodeEngineRole
E1Collection Engine뉴스 수집 (Research → Seed → Scraper)
E2Diaspora EngineCity Node 기반 다국어 뉴스 사이트 생성
E3RAG EngineAgentic GraphRAG + Knowledge API
E4Knowledge AgentDiaspora 사이트 연동 AI 에이전트
E5Journalist Engine사용자 기사 작성 지원 AI
E6Advertising Engine도메인-시티별 광고 관리 (Frappe/ERPNext)
E7Publishing Engine멀티미디어 자동 생성 및 소셜 발행
E8Distribution Engine멀티채널 뉴스 배포

The current codebase (Research → Seed → Dataset → Metadata, single Worker, Queues, R2, D1, KV) remains the foundation and maps to E1 (Collection Engine) and shared infrastructure. v6.0 adds E2–E8 as separate engines with the same naming and interface discipline. Existing docs (e.g. Seed Engine Workflow, Distributed Build and Edge Caching) continue to describe the current system; this document describes the target engine-based model.


  • Project Overview — current architecture and Tech Stack
  • docs/planning/ — E1, E2, E3 engine specs (e1-collection-engine-spec.md, e2-diaspora-engine-spec.md, e3-rag-engine-spec.md) and resource map
  • .cursor/rules/infrastructure/resource-naming.mdcnf-{engine}-{resource}-{qualifier} convention