# L-MARS: Legal Multi-Agent System with Agentic Search and Citation-Faithfulness Audit

arXiv: https://arxiv.org/abs/2509.00761 · HTML: https://mlsanity.com/paper/2509.00761/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- Placed in Social simulation and multi-agent debate, one of 16 directions in LLM agents and tool use.
- Builds on 24 of its 43 references that are in the map.
- First posted on arXiv 31 Aug 2025.

## Placed in

- LLM agents and tool use / [Social simulation and multi-agent debate](https://mlsanity.com/agents/social-simulation-and-multi-agent-debate/)

## Builds on (most cited first, 24 of 24)

- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Self-Consistency Improves Chain of Thought Reasoning in Language Models](https://mlsanity.com/paper/2203.11171/) (2022)
- [Reflexion: Language Agents with Verbal Reinforcement Learning](https://mlsanity.com/paper/2303.11366/) (2023)
- [Survey of Hallucination in Natural Language Generation](https://mlsanity.com/paper/2202.03629/) (2022)
- [AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation](https://mlsanity.com/paper/2308.08155/) (2023)
- [Language Models (Mostly) Know What They Know](https://mlsanity.com/paper/2207.05221/) (2022)
- [Selective Classification for Deep Neural Networks](https://mlsanity.com/paper/1705.08500/) (2017)
- [Enabling Large Language Models to Generate Text with Citations](https://mlsanity.com/paper/2305.14627/) (2023)
- [Search-o1: Agentic Search-Enhanced Large Reasoning Models](https://mlsanity.com/paper/2501.05366/) (2025)
- [WebThinker: Empowering Large Reasoning Models with Deep Research Capability](https://mlsanity.com/paper/2504.21776/) (2025)
- [Demonstrate-Search-Predict: Composing retrieval and language models for knowledge-intensive NLP](https://mlsanity.com/paper/2212.14024/) (2022)
- [LexGLUE: A Benchmark Dataset for Legal Language Understanding in English](https://mlsanity.com/paper/2110.00976/) (2021)
- [Agentic Retrieval-Augmented Generation: A Survey on Agentic RAG](https://mlsanity.com/paper/2501.09136/) (2025)
- [Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools](https://mlsanity.com/paper/2405.20362/) (2024)
- [Large Legal Fictions: Profiling Legal Hallucinations in Large Language Models](https://mlsanity.com/paper/2401.01301/) (2024)
- [AbstentionBench: Reasoning LLMs Fail on Unanswerable Questions](https://mlsanity.com/paper/2506.09038/) (2025)
- [LawBench: Benchmarking Legal Knowledge of Large Language Models](https://mlsanity.com/paper/2309.16289/) (2023)
- [Pile of Law: Learning Responsible Data Filtering from the Law and a 256GB Open-Source Legal Dataset](https://mlsanity.com/paper/2207.00220/) (2022)
- [Attributed Question Answering: Evaluation and Modeling for Attributed Large Language Models](https://mlsanity.com/paper/2212.08037/) (2022)
- [LEXam: Benchmarking Legal Reasoning on 340 Law Exams](https://mlsanity.com/paper/2505.12864/) (2025)
- [A Reasoning-Focused Legal Retrieval Benchmark](https://mlsanity.com/paper/2505.03970/) (2025)
- [Case law retrieval: problems, methods, challenges and evaluations in the last 20 years](https://mlsanity.com/paper/2202.07209/) (2022)

## Built on by (most cited first, 11 of 11)

- [τ-Knowledge: Evaluating Conversational Agents over Unstructured Knowledge](https://mlsanity.com/paper/2603.04370/) (2026)
- [SAGE: Benchmarking and Improving Retrieval for Deep Research Agents](https://mlsanity.com/paper/2602.05975/) (2026)
- [AIReg-Bench: Benchmarking Language Models That Assess AI Regulation Compliance](https://mlsanity.com/paper/2510.01474/) (2025)
- [LLM Agents in Law: Taxonomy, Applications, and Challenges](https://mlsanity.com/paper/2601.06216/) (2026)
- [LawThinker: A Deep Research Legal Agent in Dynamic Environments](https://mlsanity.com/paper/2602.12056/) (2026)
- [ClawTrace: Cost-Aware Tracing for LLM Agent Skill Distillation](https://mlsanity.com/paper/2604.23853/) (2026)
- [Mitigating Hallucination in Large Language Models (LLMs): An Application-Oriented Survey on RAG, Reasoning, and Agentic Systems](https://mlsanity.com/paper/2510.24476/) (2025)
- [Decompose-and-Refine: Structured Legal Question Answering with Parametric Retrieval](https://mlsanity.com/paper/2605.24454/) (2026)
- [Maat: The Agentic Legal Research Assistant for Competition Protection](https://mlsanity.com/paper/2605.27331/) (2026)
- [A Matter of TASTE: Improving Coverage and Difficulty of Agent Benchmarks](https://mlsanity.com/paper/2605.28556/) (2026)
- [LabourCrew: A Multi-Agent RAG Framework for Trustworthy Adversarial Deliberation and Statutory Reasoning over Labour Law](https://mlsanity.com/paper/2609.27814/) (2026)

## Abstract (arXiv)

Large language models are increasingly deployed for legal question answering, where evaluations typically focus on multiple-choice accuracy. This measure overlooks a common failure: whether the citation source attached to an answer exists and supports the rule the system attributes to it. We present L-MARS, an open multi-agent legal QA system with agentic search and judge-driven evidence checks, and audit it claim by claim against its cited source. Each atomic claim is labelled with a six-class taxonomy and scored with strict-ALCE under cross-provider judging, where the answerer and verifier come from different model families. On a stratified 100-question Bar Exam audit, retrieval barely moves accuracy, yet the multi-turn judge loop lifts strict citation F1 from 0.13 (naive RAG) to 0.25 and cuts the no-citation rate from 34% to 13%. We further introduce Faith-Search, a post-draft step that re-verifies and repairs unreachable citations; it drops the unreachable rate below 1% but does not improve F1 over the multi-turn loop, so we report it as a targeted reachability intervention rather than a faithfulness breakthrough. A 50-question LegalSearchQA case study confirms the picture: retrieve-then-draft pipelines saturate near 0.75 citation F1, while a single-agent web-search baseline collapses to 0.22 under external audit.