Growth-Oriented Curation for Endangered TCM Germplasm in Plant Communities: Evidence KG, Topological Structure Discovery, and World-Model Agents for an Online–Offline Living Museum

Endangered TCM germplasm conservation faces name–taxon mismatch, relocation-style ex situ failure, occurrence sensitivity, and weak exhibition–research coupling. We propose growth-oriented curation (GOC): a plant-community living-museum framework formalized as ecological niche, temporal rhythm, and community structure profiles (E,T,C); an Evidence KG with graded evidence chains; TDA and multi-scale community detection into a RuleBank; and a world-model digital twin with a grid-only Decide API and desensitization policies. Four agent roles (Evidence, Topology, World, Education) specify an online–offline multimodal system. This Version publishes methods and a frozen evaluation protocol; no measured alignment accuracy, AUC, module stability, or Top-K hit rates are claimed. A small helper (goc_ref.py) is included. Chinese and English full texts are deposited with matching section numbering (Sections 1–11 and Appendices A–C).