The paper demonstrates that a verifier used in closed‑loop agent debugging can inadvertently reveal the answer it is meant to test, rendering solver comparisons meaningless. In a study of 12 development cases, both exact minimum hitting set and a greedy method returned identical supports, and an audit showed that exact‑anchor predicates always produced the planted fault pair. The authors propose a support‑gated verification contract that requires a clean reference map and runtime evidence before an independently calibrated signal can confirm a detection, and validate this approach on 1,440 held‑out cases with a low false‑admission rate.
arXiv:2608. 14680v1 Announce Type: new Abstract: Reliability in LLM-based agentic systems is a property of the whole execution (its tool calls, model calls, guardrails, and inter-agent messages), not of the final answer alone, yet evaluating only task outcomes reveals little about how or why a run fails.
The paper introduces a claim‑safe protocol for evaluating closed‑loop AI systems, consisting of three actions: Refuse, Decompose, and Refresh. It demonstrates the protocol in a simulator with 24 policy components and 1,440 held‑out cases, showing that abstention and stable false admission rates are low while providing detailed statistical diagnostics. The approach emphasizes that evaluation results should be tied to observable support and statistical calibration rather than a single PASS/FAIL label.
FaulT-Bench is a new benchmark comprising 200 network troubleshooting scenarios across eight topologies, designed to test large‑language‑model agents on realistic, noisy user tickets that may contain false premises or incorrect fault claims. The benchmark includes 72 rewritten tickets that vary reporter confidence and detail while keeping the network state constant, allowing isolation of the impact of ticket wording on diagnosis. Evaluation of agents such as SADE, ReAct, and Claude Code shows they perform well on accurate tickets but degrade sharply on misleading or healthy‑network tickets, revealing differing failure modes and highlighting the importance of robust reasoning over unreliable input.
arXiv:2603. 13384v2 Announce Type: replace-cross Abstract: Software vulnerabilities often depend on cross-file data flow, build options, framework conventions, and runtime guards, so isolated function classifiers produce fragile and poorly calibrated warnings.
arXiv:2606. 20502v1 Announce Type: cross Abstract: Whether LLMs scoring well on vulnerability benchmarks genuinely reason about security or merely pattern-match on contaminated data remains unresolved.
The study evaluates how to best split tasks among large‑language‑model agents for cross‑border VAT determination, comparing one broad agent to configurations ranging from one to five narrow agents. Across 4,400 runs—including token‑matched and failure‑injection scenarios—the intermediate configurations achieved the highest accuracy but did not surpass the fine‑endpoint benchmark, leaving the optimal decomposition hypothesis unconfirmed. The pilot provides a preregistered heuristic for right‑sizing decomposition, along with an oracle, dataset, and analysis pipeline.
arXiv:2608. 02464v1 Announce Type: cross Abstract: LLM agents fail mid-episode -- they loop, cascade tool errors, drift off goal, fabricate results, or silently absorb corrupted content -- and the standard remedy, judging every step with a second LLM, costs more than the agent itself.
arXiv:2606. 10241v1 Announce Type: new Abstract: Autonomous improvement loops are hard to trust because the improvement process is usually external scaffolding bolted onto the agent: failures go unlogged, diagnoses cannot be replayed, and promote-or-discard decisions land in a side database rather than the agent's own history.
arXiv:2606. 19808v1 Announce Type: new Abstract: Test-time reasoning is increasingly used as a serving-time control knob, but extra reasoning is not uniformly valuable: it can repair failed attempts, waste compute on already-correct answers, or introduce harmful answer changes.
The paper evaluates hard‑gate candidacy for validators in a deployed generative‑agent system by measuring how well each validator’s firing separates successful from failed builds. Across 13 validators and thousands of builds, only a few checks show statistically significant separation, while many fail to distinguish or never fire. The study highlights that skipped checks are recorded as passes, limiting detectable failure rates and underscoring the need for clearer evaluation records.
arXiv:2608. 14863v1 Announce Type: cross Abstract: LLM-based coding agents have advanced rapidly on single-process SWE tasks, with frontier models now clustering in the high-70s on SWE-bench Verified.
arXiv:2605. 27784v2 Announce Type: replace Abstract: LLM agents are governed by long-lived prompt policies, where individually reasonable stand- ing rules can jointly govern the same pre- generation state.