VISA: A Structured Description Protocol for Agent-Based Simulation Models Towards Machine Reproducibility

Agent-based models (ABMs) rely on simple, explicit and reproducible rules for individual decision making, while complex collective behavior emerges from interactions among agents. Recent advances in large language models (LLMs) make it tempting to replace, enrich, or perturb these rules with LLM-based agentic capabilities. However, this raises a methodological question: how does introducing LLM-driven decisions affect the reliability, computational cost, and behavior of ABM simulations? We investigate this for Mesa ABM models, a popular Python library for ABMs, analyzed by statistical model checking. Building on Mesa's integration with the statistical model checker MultiVeStA, we extend the classical Schelling segregation model with a hybrid population: ordinary agents classify neighbors using the standard symbolic rule, while one agent delegates this task to an LLM through tool calls. The LLM-enabled agent receives natural-language descriptions of neighboring agents and invokes tools that increment counters of similar/different neighbors; these counters determine its happiness according to the original Schelling dynamics. This provides a minimal but controlled setting where the semantic, operational, and computational behavior of LLM-based decisions can be studied inside an otherwise standard ABM. We report preliminary experiments with locally served LLMs of different sizes, showing that smaller models may fail simple semantic classification experiments or become operationally unusable during repeated tool-call generation, while larger tested models pass these preliminary checks. We discuss how statistical model checking can estimate classical ABM observables and quantify the impact of introducing agentic LLM components into simulation models.

Agent-Based Modelling (ABM) relies on clear documentation to ensure credibility and transparency. Although standards exist for documenting models (e.g. ODD), processes (e.g. TRACE, EABSS), and data use (e.g. RAT-RS), their adoption remains limited due to the effort required to produce documentation that is often treated as supplementary. This paper explores the use of Large Language Models (LLMs) to facilitate and partially automate such processes. We conduct a feasibility study focusing on the underused Rigour and Transparency Reporting Standard (RAT-RS), using four LLMs to extract reports from a published ABM paper. We assess consistency and performance across question types, finding that LLMs generate coherent outputs and perform more reliably on descriptive than on explanatory or evaluative tasks. While LLMs can improve reporting quality and consistency, they also exhibit notable limitations. We identify practical heuristics for when LLM-assisted documentation is reliable and when human oversight is needed and call for systematic community-level exploration to enhance rigour and adoption in ABM reporting.

LLM-agent simulations make natural-language social scenarios easy to instantiate, but their outputs can be overread as predictions and are often difficult to compare with explicit social dynamics. We present AgoraSim, a hybrid agent-based modeling framework for scenario-oriented social reaction analysis. AgoraSim resolves textual or multimodal artifacts into editable ABM configurations, runs ratio-controlled populations that mix LLM, vision-language, custom-endpoint, random, and classical agents, and compares the same scenario against matched classical reference dynamics. All agents emit a shared structured decision object, enabling common action spaces, interaction protocols, metrics, and audit records. Exposed through a local UI, Python SDK/CLI, and REST API, AgoraSim helps users inspect scenario trajectories, compare modeling assumptions, and identify cases that warrant empirical validation.