Beyond Formal Semantics for Capabilities and Skills: Model Context Protocol in Manufacturing
Luis Miguel Vieira da Silva, Aljosha Köcher, Felix Gehlhoff
TL;DR
The paper tackles the burden of formal capability/skill modeling in dynamic manufacturing by introducing the Model Context Protocol (MCP), a lightweight open standard that allows LLM-based agents to discover, describe, and invoke external tool functions at runtime. It presents an MCP architecture with direct and gateway servers to expose capabilities and illustrates how natural-language capability descriptions can replace explicit models while maintaining actionable interfaces. Through a proof-of-concept evaluation in a lab manufacturing setup, the work demonstrates planning and executing multi-step tasks via MCP-exposed tools without handcrafted control flows, while acknowledging reliability and robustness challenges. The findings suggest that model-free, tool-based integration can reduce integration barriers and enhance reconfigurability in industrial production, paving the way for broader, tool-driven LLM automation in manufacturing.
Abstract
Explicit modeling of capabilities and skills -- whether based on ontologies, Asset Administration Shells, or other technologies -- requires considerable manual effort and often results in representations that are not easily accessible to Large Language Models (LLMs). In this work-in-progress paper, we present an alternative approach based on the recently introduced Model Context Protocol (MCP). MCP allows systems to expose functionality through a standardized interface that is directly consumable by LLM-based agents. We conduct a prototypical evaluation on a laboratory-scale manufacturing system, where resource functions are made available via MCP. A general-purpose LLM is then tasked with planning and executing a multi-step process, including constraint handling and the invocation of resource functions via MCP. The results indicate that such an approach can enable flexible industrial automation without relying on explicit semantic models. This work lays the basis for further exploration of external tool integration in LLM-driven production systems.