Formal Semantics for Agentic Tool Protocols: A Process Calculus Approach
A research paper introduced a process calculus formalization for agent-tool integration protocols, specifically comparing Schema-Guided Dialogue (SGD) and the Model Context Protocol (MCP). The study found that while MCP can be mapped to SGD, the reverse mapping is partial, indicating limitations in MCP's expressivity. To achieve full equivalence, the paper proposed five principles for schema quality, which were formalized into an extended protocol called MCP+. AI
IMPACT Provides a formal foundation for verified agent systems, potentially improving the safety and reliability of LLM agents.