arXiv:2606. 09863v1 Announce Type: new Abstract: LLM agents can fail silently by asserting task completion when the environment state shows otherwise.
arXiv:2610.17747v1 Announce Type: cross Abstract: Failure-attribution benchmarks ask an LLM judge which step or agent caused a failure. A high score should mean the judge checked the faulty step agai...
The paper investigates whether human or LLM‑judge‑identified ‘first mistake’ points truly indicate failure causes in LLM agents. By reconstructing intermediate states and computing a prefix value V(h_t) for each step, the authors identify a certified point of no return (PNR) and compare it to judge‑labeled steps. Their experiments on 150 SWE‑bench tasks show that judge labels often miss the decisive step, while restarting at the measured PNR improves success rates from 0.44 to 0.57.
arXiv:2610.17633v1 Announce Type: new Abstract: Work on debugging LLM agents treats the "first mistake" marked by a human or LLM judge as ground truth for where a trajectory failed. We ask whether su...
arXiv:2608.22808v2 Announce Type: replace Abstract: When can an agent failure be caught? An audit is usually limited by the record rather than by the method. CatchBench therefore puts one auditor's q...
arXiv:2606. 01365v1 Announce Type: new Abstract: Tool-using multi-agent large language model (LLM) systems spend computation through model tokens, tool calls, retries, and code execution before producing an answer.
The paper introduces TRACE, a protocol designed to diagnose whether changes in verifier scores for large language model agents reflect actual changes in agent behavior or merely alterations in the evaluation process. TRACE works by applying targeted changes to evaluation components, running paired experiments, and re‑scoring unchanged trajectories to isolate the source of score variation. Experiments on synthetic tasks and public benchmark tasks demonstrate that seemingly significant score shifts can often be attributed to evaluation artifacts, while TRACE can also detect genuine behavioral changes.
The paper introduces Traverse, a benchmark of 2,518 agent trajectories and 6,967 annotated mistakes across software engineering, computer use, and science tasks, revealing that failures often go unrecovered and can cause irreversible harm before a run is deemed successful. It shows that human judges struggle to detect the first mistake in most runs, while a 4‑billion‑parameter verifier called Scout can locate failures more effectively and improve task success when used to select among candidate runs. The study demonstrates that making failure detection inexpensive and reliable can enable long‑horizon agents to learn from their own mistakes and increase trustworthiness in autonomous AI.
arXiv:2607. 09996v1 Announce Type: new Abstract: Automated failure attribution uses LLMs to identify where and why agentic systems fail.
The paper introduces SOAP, a method for attributing failures in large language model agents without requiring step-level labels or a separate set of successful trajectories. SOAP builds a spectral reference space from LLM representations to assign base error scores to each step, then uses attention-based soft dependencies to propagate downstream error signals back to their likely upstream causes. Experiments demonstrate that SOAP outperforms existing baselines in accurately identifying decisive steps responsible for task failures.
arXiv:2602. 11619v2 Announce Type: replace Abstract: Running the same LLM agent on identical inputs yields 2.
arXiv:2610.03448v1 Announce Type: cross Abstract: LLM agents increasingly screen tool outputs with small prompt-injection detectors, and teams choose among detectors by their scores on public benchma...