# AgentModernize: Preserving Business Logic in Legacy Modernization with Multi-Agent LLMs and Behavioral Specification Graphs

arXiv: https://arxiv.org/abs/2605.17535 · HTML: https://mlsanity.com/paper/2605.17535/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Software engineering coding agents, one of 16 directions in LLM agents and tool use.
- Builds on 13 of its 31 references that are in the map.
- First posted on arXiv 17 May 2026.

## Placed in

- LLM agents and tool use / [Software engineering coding agents](https://mlsanity.com/agents/software-engineering-coding-agents/)

## Builds on (most cited first, 13 of 13)

- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [Reflexion: Language Agents with Verbal Reinforcement Learning](https://mlsanity.com/paper/2303.11366/) (2023)
- [Self-Refine: Iterative Refinement with Self-Feedback](https://mlsanity.com/paper/2303.17651/) (2023)
- [MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework](https://mlsanity.com/paper/2308.00352/) (2023)
- [SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering](https://mlsanity.com/paper/2405.15793/) (2024)
- [OpenHands: An Open Platform for AI Software Developers as Generalist Agents](https://mlsanity.com/paper/2407.16741/) (2024)
- [ChatDev: Communicative Agents for Software Development](https://mlsanity.com/paper/2307.07924/) (2023)
- [StarCoder: may the source be with you!](https://mlsanity.com/paper/2305.06161/) (2023)
- [Large Language Models for Software Engineering: A Systematic Literature Review](https://mlsanity.com/paper/2308.10620/) (2023)
- [Automating Code Review Activities by Large-Scale Pre-training](https://mlsanity.com/paper/2203.09095/) (2022)
- [Less Training, More Repairing Please: Revisiting Automated Program Repair via Zero-shot Learning](https://mlsanity.com/paper/2207.08281/) (2022)
- [Metamorphic Testing: A New Approach for Generating Next Test Cases](https://mlsanity.com/paper/2002.12543/) (2020)
- [Large Language Models for Code Analysis: Do LLMs Really Do Their Job?](https://mlsanity.com/paper/2310.12357/) (2023)

## Built on by (most cited first, 1 of 1)

- [From Code to Requirements: Agentic Reverse Engineering of Business Rules at Enterprise Scale](https://mlsanity.com/paper/2609.22719/) (2026)

## Abstract (arXiv)

Legacy modernization breaks business logic more often than most teams expect. Most tools and LLM-based approaches treat modernization as syntax translation: convert COBOL to Java, swap PL/SQL for Python, ship it. Implicit rules, edge-case handling, and cross-module constraints that keep production systems running are lost, and nobody notices until something fails in production. We present AgentModernize, a multi-agent framework that treats modernization as a behavioral preservation problem. Four agents handle extraction, specification, code generation, and validation. The key intermediate artifact, a Behavioral Specification Graph (BSG), forces extracted business logic to be explicit and inspectable before any code is generated. We evaluated on LegacyModernize-8, eight synthetic scenarios spanning telecom and banking, under a fair protocol where each method's tests are generated from its own API surface (3 trials, temperature 0.0). With GPT-4o-mini, AgentModernize with feedback achieves 23.0% mean BER (non-zero on 5/8 scenarios, up to 53.3%), while AgentModernize without feedback reaches 23.8%. SP-LLM scores 12.4% (non-zero on 2/8) and CoT-LLM 4.5% (2/8). No single method dominates all scenarios. The feedback loop is decisive for scenarios requiring iterative correction (S4, S6, S7) but can regress code in others (S2, S8). The BSG captures 92.3% of gold-standard rules with 90.2% precision; the bottleneck is code generation, not extraction. A cross-model study (GPT-4o, GPT-5.3-codex) reveals the pipeline's benefit is inversely correlated with model capability: stronger models achieve higher BER with single-prompt methods than with the pipeline. For regulated industries, the pipeline's traceable artifacts (business rule inventory, BSG, equivalence reports) provide an audit trail that no single-prompt approach can match.