mlsanityData to 14 Oct 2026

RAFFLES: Reasoning-based Attribution of Faults for LLM Systems

arXiv 2509.06822 ↗First posted 8 Sep 2025cs.AI cs.CLcited by 21

On the map: LLM reasoning → Reasoning for agents, planning and code

Abstract (arXiv)

The advent of complex, interconnected long-horizon LLM systems has made it incredibly tricky to identify where and when these systems break down. Evaluation capabilities that currently exist today are limited in that they often focus on simple metrics, end-to-end outcomes, and are dependent on the perspectives of humans. In order to match the increasing complexity of these many component systems, evaluation frameworks must also be able to reason, probe, iterate, and understand the nuanced logic passing through these systems. In this paper, we present RAFFLES, an offline evaluation architecture that incorporates iterative reasoning. Specifically, RAFFLES operates as an iterative, multi-component pipeline, using a central Judge to systematically identify faults and a set of specialized Evaluators to assess the quality of the candidate faults as well as rationales of the Judge. We evaluated RAFFLES with several benchmarks - the Who&When dataset to identify step-level faults in multi-agent systems and the ReasonEval datasets to diagnose step-level mathematical reasoning errors. RAFFLES outperforms strong baselines, achieving an accuracy of over 20% and 50% on the Who&When Hand-Crafted and Algorithmically-Generated datasets, and over 80% on the ReasonEval datasets. These results demonstrate a key step towards introducing automated fault detection for autonomous systems over labor-intensive manual review.

Key facts. Cited by 21 papers in the map (data through 14 Oct 2026). Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning. Builds on 44 of its 51 references that are in the map. First posted on arXiv 8 Sep 2025.

Builds on

44 of 51 references.

Closest ancestors · 1

Same direction, newest first

  1. 2025TRAIL: Trace Reasoning and Agentic Issue Localization13 May

Other references · 43

  1. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  2. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,695
  3. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
  4. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,944
  5. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,708
  6. 2022Large Language Models are Zero-Shot Reasonerscited by 4,374
  7. 2023Reflexion: Language Agents with Verbal Reinforcement Learningcited by 4,276
  8. 2023Toolformer: Language Models Can Teach Themselves to Use Toolscited by 4,182
  9. 2023Let's Verify Step by Stepcited by 3,789
  10. 2023Lost in the Middle: How Language Models Use Long Contextscited by 3,455
  11. 2023Self-Refine: Iterative Refinement with Self-Feedbackcited by 3,442
  12. 2023SWE-bench: Can Language Models Resolve Real-World GitHub Issues?cited by 3,373
31 more
  1. 2023Tree of Thoughts: Deliberate Problem Solving with Large Language Modelscited by 3,211
  2. 2023G-Eval: NLG Evaluation using GPT-4 with Better Human Alignmentcited by 2,276
  3. 2022Least-to-Most Prompting Enables Complex Reasoning in Large Language Modelscited by 1,407
  4. 2025gpt-oss-120b & gpt-oss-20b Model Cardcited by 1,386
  5. 2023GAIA: a benchmark for General AI Assistantscited by 1,128
  6. 2023Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMscited by 974
  7. 2023Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedbackcited by 812
  8. 2022Teaching Models to Express Their Uncertainty in Wordscited by 751
  9. 2023ChatEval: Towards Better LLM-based Evaluators through Multi-Agent Debatecited by 742
  10. 2025Why Do Multi-Agent LLM Systems Fail?cited by 604
  11. 2022Decomposed Prompting: A Modular Approach for Solving Complex Taskscited by 558
  12. 2025The Lessons of Developing Process Reward Models in Mathematical Reasoningcited by 415
  13. 2024TravelPlanner: A Benchmark for Real-World Planning with Language Agentscited by 392
  14. 2023GPTScore: Evaluate as You Desirecited by 361
  15. 2024AgentBoard: An Analytical Evaluation Board of Multi-turn LLM Agentscited by 339
  16. 2024Replacing Judges with Juries: Evaluating LLM Generations with a Panel of Diverse Modelscited by 326
  17. 2024ProcessBench: Identifying Process Errors in Mathematical Reasoningcited by 292
  18. 2024Agent-as-a-Judge: Evaluate Agents with Agentscited by 238
  19. 2025Which Agent Causes Task Failures and When? On Automated Failure Attribution of LLM Multi-Agent Systemscited by 185
  20. 2023SelfCheck: Using LLMs to Zero-Shot Check Their Own Step-by-Step Reasoningcited by 159
  21. 2024AssistantBench: Can Web Agents Solve Realistic and Time-Consuming Tasks?cited by 158
  22. 2024Autonomous Evaluation and Refinement of Digital Agentscited by 137
  23. 2025PRMBench: A Fine-grained and Challenging Benchmark for Process-Level Reward Modelscited by 99
  24. 2025Learning to Plan & Reason for Evaluation with Thinking-LLM-as-a-Judgecited by 78
  25. 2025Interactive Debugging and Steering of Multi-Agent AI Systemscited by 68
  26. 2024Inference Scaling for Long-Context Retrieval Augmented Generationcited by 65
  27. 2025R-PRM: Reasoning-Driven Process Reward Modelingcited by 38
  28. 2025MCPEval: Automatic MCP-based Deep Evaluation for AI Agent Modelscited by 36
  29. 2025Does Context Matter? ContextualJudgeBench for Evaluating LLM-based Judges in Contextual Settingscited by 24
  30. 2025SHIELDA: Structured Handling of Exceptions in LLM-Driven Agentic Workflowscited by 13
  31. 2025How Well Can Reasoning Models Identify and Recover from Unhelpful Thoughts?cited by 11

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 21.

21 papers.

Sort:

By area:

  1. 2026Who&When Pro: Can LLMs Really Attribute Failures in AI Agents?cited by 18
  2. 2026Your Agents Are Aging Too: Agent Lifespan Engineering for Deployed Systemscited by 9
  3. 2026When Agents Fail: A Comprehensive Study of Bugs in LLM Agents with Automated Labelingcited by 6
  4. 2026Efficient Failure Management for Multi-Agent Systems with Reasoning Trace Representationcited by 6
  5. 2026Conformal Agent Error Attributioncited by 4
  6. 2026CUADebug: Diagnosing and Repairing Computer-Use Agent Failurescited by 4
  7. 2026StepFinder: A Temporal Semantic Framework for Failure Attribution in Multi-Agent Systemscited by 3
  8. 2026EDGE: Error Dependency Graph-Guided Multi-Error Attribution in Multi-Agent LLM Systemscited by 3
  9. 2026QUIVER: A Formal Framework for Quantifying Perturbation Propagation and Bifurcation in Compound AI Systemscited by 2
  10. 2026LongRCA Bench: Root-Cause Localization in Long-Horizon Agent Trajectoriescited by 2
  11. 2026Themis: Training Robust Multilingual Code Reward Models for Flexible Multi-Criteria Scoringcited by 1
  12. 2026SAFARI: Scaling Long Horizon Agentic Fault Attribution via Active Investigationcited by 1
9 more
  1. 2026Autonomous Repair for Multi-Agent Systems via Monte-Carlo Tree Searchcited by 1
  2. 2026ASCon: A Direction-Aware Reciprocal Agent--Step Contextualization Model for Failure Attribution in Multi-Agent Systemscited by 1
  3. 2026Adaptive Influence Graphs for Failure Attribution in Multi-Agent Systemscited by 1
  4. 2026Towards a Science of Collective AI: LLM-based Multi-Agent Systems Need a Transition from Blind Trial-and-Error to Rigorous Sciencecited by 0
  5. 2026Diagnosis-Driven Automatic Repair for Agentic Workflow via Symbolic Inferencecited by 0
  6. 2026Localizing Emergent Failures in Agentic AI: Recovering Minimal Repair Families via Counterfactual Replaycited by 0
  7. 2026Diagnosing with Insights: Structured Analysis of Agent Failures via Behavioral Abstractionscited by 0
  8. 2026Can Failed Trajectories Reveal Where Agents Go Wrong? Agentic Failure Attribution Without Step-Level Labelscited by 0
  9. 2026Structural Baselines Matter for Agent Failure Detection and Attributioncited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026A Multi-Dataset Benchmark for Evaluating LLM Agents in Microservice Failure Diagnosis
  2. 2026MAS-FIRE: Fault Injection and Reliability Evaluation for LLM-Based Multi-Agent Systems
  3. 2026ReFlect: An Effective Harness System for Complex Long-Horizon LLM Reasoning
  4. 2025ReasonBENCH: Benchmarking the (In)Stability of LLM Reasoning
  5. 2026AgentFixer: From Failure Detection to Fix Recommendations in LLM Agentic Systems
  6. 2026Seeing the Whole Elephant: A Benchmark for Failure Attribution in LLM-based Multi-Agent Systems
  7. 2025Breakpoint: Scalable evaluation of system-level reasoning in LLM code agents
  8. 2026DIANOIA: Diagnostic Decomposition and Joint Optimization for Multi-Agent Reasoning