38 pre-built components organized into 5 tiers. Each component is reusable across multiple use cases. Team: please review and flag anything missing, incorrect, or misplaced.
For each connector: What API version? What endpoints? What auth method (OAuth, session, API key)? Any existing integration code or Postman collections we can reuse? Are there systems we're missing?
For each skill: Is the description accurate? Are we using the right methodology (e.g., 5-Why vs Fishbone for RCA)? What risk matrix do clients use? Can you share a sample SOP, audit report, or CAPA form so we can write accurate prompt templates?
For each agent: Is the purpose description accurate? Are the right tools/skills attached? Are we missing any agents that clients frequently need? Should any agents be split or merged?
For each flow: Is the step sequence correct? Are we missing steps? Which flows should we build first based on client demand? Are there flows we haven't thought of?
System ──GOVERNED_BY──→ SOPSOP ──DEFINES──→ ProcedureProcedure ──REQUIRES_ROLE──→ RoleUser ──HAS_ROLE──→ RoleUser ──AUTHORIZED_ON──→ SystemTicket ──RAISED_FOR──→ SystemTicket ──APPROVED_BY──→ User
Product ──HAS_BATCH──→ BatchDeviation ──AFFECTS──→ BatchCAPA ──ADDRESSES──→ DeviationComplaint ──ABOUT──→ ProductSupplier ──SUPPLIES──→ ProductTraining ──REQUIRED_BY──→ RoleTraining ──COMPLETED_BY──→ User
Ontology review: Are the entity types correct? Are we using the right terminology? (CAPA vs Corrective Action?) Are relationships accurate? Any entities or relationships missing? Does CG have an existing data model we should align with?