# Where Do Deep-Research Agents Go Wrong? Span-Level Error Localization in Agent Trajectories

arXiv: https://arxiv.org/abs/2606.02060 · HTML: https://mlsanity.com/paper/2606.02060/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Placed in Web, search, and research agents, one of 16 directions in LLM agents and tool use.
- Builds on 24 of its 47 references that are in the map.
- First posted on arXiv 1 Jun 2026.

## Placed in

- LLM agents and tool use / [Web, search, and research agents](https://mlsanity.com/agents/web-search-and-research-agents/)

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

- [Let's Verify Step by Step](https://mlsanity.com/paper/2305.20050/) (2023)
- [ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIs](https://mlsanity.com/paper/2307.16789/) (2023)
- [FEVER: a large-scale dataset for Fact Extraction and VERification](https://mlsanity.com/paper/1803.05355/) (2018)
- [DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models](https://mlsanity.com/paper/2512.02556/) (2025)
- [BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agents](https://mlsanity.com/paper/2504.12516/) (2025)
- [Why Do Multi-Agent LLM Systems Fail?](https://mlsanity.com/paper/2503.13657/) (2025)
- [ProcessBench: Identifying Process Errors in Mathematical Reasoning](https://mlsanity.com/paper/2412.06559/) (2024)
- [DeepResearch Bench: A Comprehensive Benchmark for Deep Research Agents](https://mlsanity.com/paper/2506.11763/) (2025)
- [Which Agent Causes Task Failures and When? On Automated Failure Attribution of LLM Multi-Agent Systems](https://mlsanity.com/paper/2505.00212/) (2025)
- [AVeriTeC: A Dataset for Real-world Claim Verification with Evidence from the Web](https://mlsanity.com/paper/2305.13117/) (2023)
- [VisualPRM: An Effective Process Reward Model for Multimodal Reasoning](https://mlsanity.com/paper/2503.10291/) (2025)
- [xbench: Tracking Agents Productivity Scaling with Profession-Aligned Real-World Evaluations](https://mlsanity.com/paper/2506.13651/) (2025)
- [TRAIL: Trace Reasoning and Agentic Issue Localization](https://mlsanity.com/paper/2505.08638/) (2025)
- [PRMBench: A Fine-grained and Challenging Benchmark for Process-Level Reward Models](https://mlsanity.com/paper/2501.03124/) (2025)
- [AgentRx: Diagnosing AI Agent Failures from Execution Trajectories](https://mlsanity.com/paper/2602.02475/) (2026)
- [DeepResearchGym: A Free, Transparent, and Reproducible Evaluation Sandbox for Deep Research](https://mlsanity.com/paper/2505.19253/) (2025)
- [Can Large Language Models Detect Errors in Long Chain-of-Thought Reasoning?](https://mlsanity.com/paper/2502.19361/) (2025)
- [LiveResearchBench: A Live Benchmark for User-Centric Deep Research in the Wild](https://mlsanity.com/paper/2510.14240/) (2025)
- [AgentProcessBench: Diagnosing Step-Level Process Quality in Tool-Using Agents](https://mlsanity.com/paper/2603.14465/) (2026)
- [Seeing the Whole Elephant: A Benchmark for Failure Attribution in LLM-based Multi-Agent Systems](https://mlsanity.com/paper/2604.22708/) (2026)
- [Beyond the Final Answer: Evaluating the Reasoning Trajectories of Tool-Augmented Agents](https://mlsanity.com/paper/2510.02837/) (2025)
- [DRBench: A Realistic Benchmark for Enterprise Deep Research](https://mlsanity.com/paper/2510.00172/) (2025)
- [CodeTracer: Towards Traceable Agent States](https://mlsanity.com/paper/2604.11641/) (2026)
- [Agentic-MME: What Agentic Capability Really Brings to Multimodal Intelligence?](https://mlsanity.com/paper/2604.03016/) (2026)

## Built on by (most cited first, 6 of 6)

- [Autonomous Research Agents: A Survey of AI Scientists and the Verification Gap](https://mlsanity.com/paper/2608.05179/) (2026)
- [Agent Error Dataset: Scaling 50,000 Error--Diagnosis Pairs for Failure Analysis and Error-Aware Post-Training](https://mlsanity.com/paper/2609.40111/) (2026)
- [Mint-Agent: Introducing Finance-Native Agentic Foundation Models](https://mlsanity.com/paper/2608.16386/) (2026)
- [SearchAtlas: Analyzing Agentic Search Strategies via Evidential Query Graphs](https://mlsanity.com/paper/2609.10901/) (2026)
- [AgentPProf: Semantic Profiler for Long Horizon AI Agents](https://mlsanity.com/paper/2609.20301/) (2026)
- [Auditing Agent Actions through Query-Conditioned Attribution](https://mlsanity.com/paper/2609.33676/) (2026)

## Abstract (arXiv)

Deep-research agents solve tasks through long trajectories of search, tool use, evidence inspection, and answer synthesis. Evaluation based on final answers shows whether an agent succeeds, but not which parts of the trajectory make the answer unreliable. We study span-level error localization for deep-research agents. We collect 2,790 real trajectories from two agent frameworks, three backbone models, and three benchmarks, convert raw logs into semantic spans, and annotate harmful error spans through LLM-assisted expert review. From these annotations, we build TELBench, a 1,000-instance benchmark for identifying error spans among normal exploration, failed searches, tentative hypotheses, and harmless noise. We further propose DRIFT, a claim-centric auditing framework that tracks agent claims, checks their support in trajectory evidence, and marks spans where unsupported or conflicting claims affect the answer path. Experiments across model families and auditing frameworks show that DRIFT improves span-level error localization and first-error accuracy by up to 30 percentage points. Our work provides a process-level view of reliability in deep-research agents.