Benchmark for Planning and Control with Large Language Model Agents: Blocksworld with Model Context Protocol

TL;DR

This paper introduces a standardized benchmark and simulation environment for evaluating Large Language Model agents in Blocksworld planning tasks, enabling systematic comparison of different approaches.

Contribution

It presents a novel benchmark with an integrated Model Context Protocol for flexible, standardized evaluation of LLM-based planning agents in complex environments.

Findings

Benchmark covers five complexity categories.

Single-agent implementation validates benchmark's applicability.

Provides quantitative metrics for comparing LLM planning approaches.

Abstract

Industrial automation increasingly requires flexible control strategies that can adapt to changing tasks and environments. Agents based on Large Language Models (LLMs) offer potential for such adaptive planning and execution but lack standardized benchmarks for systematic comparison. We introduce a benchmark with an executable simulation environment representing the Blocksworld problem providing five complexity categories. By integrating the Model Context Protocol (MCP) as a standardized tool interface, diverse agent architectures can be connected to and evaluated against the benchmark without implementation-specific modifications. A single-agent implementation demonstrates the benchmark's applicability, establishing quantitative metrics for comparison of LLM-based planning and execution approaches.