{"paper":{"title":"AgentModernize: Preserving Business Logic in Legacy Modernization with Multi-Agent LLMs and Behavioral Specification Graphs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A multi-agent framework with an explicit behavioral graph preserves business logic during legacy modernization where direct LLM translation loses it.","cross_cats":[],"primary_cat":"cs.SE","authors_text":"Marnim Galib, Sheikh Nazib Ahmed","submitted_at":"2026-05-17T16:39:48Z","abstract_excerpt":"Legacy modernization breaks business logic. Most tools and LLM-based approaches treat modernization as syntax translation, losing implicit rules, edge-case handling, and cross-module constraints. We present AgentModernize, a multi-agent framework that treats modernization as a behavioral preservation problem. Four specialized 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 LegacyModerniz"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Full AgentModernize with feedback was the only configuration with non-zero mean BER under every backbone. SP-LLM and CoT-LLM scored 0.0% on every scenario, on every backbone. The BSG captures 91.2% of gold-standard rules, confirming that the bottleneck is code generation, not extraction.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The gold-standard tests used in the LegacyModernize-8 evaluation comprehensively cover all implicit business logic, edge cases, and cross-module constraints present in the original legacy systems (evaluation section).","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A multi-agent LLM framework with Behavioral Specification Graphs preserves business logic in legacy modernization, achieving non-zero mean BER on all tested scenarios where baseline LLM approaches scored zero.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A multi-agent framework with an explicit behavioral graph preserves business logic during legacy modernization where direct LLM translation loses it.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"f4741a64cc14d17acd68b0ace93e4b0672aedb68ea37d7e9b705225682fb3fa2"},"source":{"id":"2605.17535","kind":"arxiv","version":1},"verdict":{"id":"b742e2a1-df28-4fea-a967-c7f6a4475557","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T22:35:52.892603Z","strongest_claim":"Full AgentModernize with feedback was the only configuration with non-zero mean BER under every backbone. SP-LLM and CoT-LLM scored 0.0% on every scenario, on every backbone. The BSG captures 91.2% of gold-standard rules, confirming that the bottleneck is code generation, not extraction.","one_line_summary":"A multi-agent LLM framework with Behavioral Specification Graphs preserves business logic in legacy modernization, achieving non-zero mean BER on all tested scenarios where baseline LLM approaches scored zero.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The gold-standard tests used in the LegacyModernize-8 evaluation comprehensively cover all implicit business logic, edge cases, and cross-module constraints present in the original legacy systems (evaluation section).","pith_extraction_headline":"A multi-agent framework with an explicit behavioral graph preserves business logic during legacy modernization where direct LLM translation loses it."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.17535/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_title_agreement","ran_at":"2026-05-19T23:01:19.497020Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T22:41:16.671849Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T21:33:23.616696Z","status":"skipped","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T21:21:57.553949Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"8f74494dddd3bf158b228fc453a7986ed0eb7f3cc4609690ba72e38ebc63df7d"},"references":{"count":26,"sample":[{"doi":"","year":2000,"title":"A survey of legacy system modernization approaches,","work_id":"5a7225bc-59b1-401a-88af-3beb8fb19d6e","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1016/j.infsof.2011.08.002","year":2012,"title":"A systematic review of software architecture evolution research,","work_id":"94d2d7ed-0232-49d0-ad75-27fe0bf470f2","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1109/wcre.1998.723185","year":1998,"title":"Requirements for integrating software architecture and reengineering models: CORUM II,","work_id":"37ba8f60-1002-4e7e-a357-216d6e5ce068","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1109/ms.2004.23","year":2004,"title":"Model-driven reverse engi- neering,","work_id":"4422e0d2-f1c2-4617-a3ba-6b72476ca4a1","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"Evaluating Large Language Models Trained on Code","work_id":"042493e9-b26f-4b4e-bbde-382072ca9b08","ref_index":5,"cited_arxiv_id":"2107.03374","is_internal_anchor":true}],"resolved_work":26,"snapshot_sha256":"dbe712b6162bb73ddc83b92145660a30485ee7de96c0e9cc9df0c9d84dd870ec","internal_anchors":4},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}