Bridging Real and Virtual Worlds: A Blockchain-IoT-Agent Framework for the Metaverse

The convergence of Internet of Things (IoT), blockchain, and large language model (LLM) agents enables new metaverse applications where digital twins can reason about and act upon real-world assets. This paper introduces DAMIF, a Decentralized Agent-based Metaverse Integration Framework, that bridges physical sensors, smart contracts, and intelligent agents through a layered architecture comprising memory, communication, intelligence, and user interface components. A key feature is the use of the emerging Model Context Protocol (MCP) to standardize agent-to-iot, agent-to-blockchain, and agent-to-tool interactions. We implement two proof-of-concept scenarios: a smart-building concierge that authenticates and grants access to visitors, and a household energy predictor that predicts and optimizes power flows among PV, EV, and battery subsystems. Results demonstrate architectural feasibility and highlight practical challenges such as tool-call reliability and prompt engineering. DAMIF provides a flexible, protocol-driven foundation for intelligent cyber-physical systems, paving the way for modular and scalable agent-based automation in the industrial metaverse.