MCP-Cosmos: World Model-Augmented Agents for Complex Task Execution in MCP Environments

2026-05-09

AI Summary

This work addresses the limited long-term predictive capability of existing Model Context Protocol (MCP) agents, which often results in a disconnect between task planning and real-time execution. To bridge this gap, the authors propose the “Bring Your Own World Model” (BYOWM) strategy, which integrates a generative world model into the MCP ecosystem. By simulating state transitions and optimizing plans within a latent space, BYOWM enables closed-loop coordination between planning and execution. The approach synergistically combines MCP, generative world models, and ReAct/SPIRAL agent frameworks to support multi-model collaborative reasoning. Evaluated on over twenty tasks in MCP-Bench, the method significantly improves tool invocation success rates and parameter accuracy, while also introducing novel evaluation metrics such as Execution Quality.

Abstract

The Model Context Protocol (MCP) has unified the interface between Large Language Models (LLMs) and external tools, yet a fundamental gap remains in how agents conceptualize the environments within which they operate. Current paradigms are bifurcated: Task-level planning often ignores execution-time dynamics, while reactive execution lacks long-horizon foresight. We present MCP-Cosmos, a framework that infuses generative World Models (WM) into the MCP ecosystem to enable predictive task automation. By unifying three disparate technologies, namely MCP, World Model, and Agent, we demonstrate that a "Bring Your Own World Model" (BYOWM) strategy allows agents to simulate state transitions and refine plans in a latent space before execution. We conducted experiments using two strategies, namely ReAct and SPIRAL with 2 planning models and 3 representative world models over 20+ MCP-Bench tasks. We observed improvements in Agent's environment interaction KPI such as tool success rate and tool parameter accuracy. The framework also offers new metrics such as Execution Quality to generate new insights about the effectiveness of world models compared to baseline.

Problem

• Model Context Protocol
• World Model
• Task Execution
• Agent Planning
• Environment Modeling

Innovation

• World Model
• Model Context Protocol
• Predictive Task Automation
• BYOWM
• Agent Planning