arXiv:2608.29685v1 Announce Type: new Abstract: Early failure prediction is important for long-horizon agents, as it enables timely intervention and can reduce inference and tool-use costs. Uncertain...
arXiv:2602. 11619v2 Announce Type: replace Abstract: Running the same LLM agent on identical inputs yields 2.
LEASH is a training‑free framework that enables uncertainty‑aware escalation from a weak to a strong computer‑use agent. It uses latent uncertainty signals from the weak agent’s hidden states to decide when to hand off control, and a shadow observer to maintain oversight of the strong black‑box agent. Experiments on OSWorld show LEASH improves task success over standalone weak agents and outperforms random escalation, achieving comparable performance to Claude Opus 5 while cutting inference cost by 19.2%.
arXiv:2610.18614v1 Announce Type: new Abstract: Controlling an acting LLM agent means deciding, at each step of a long-horizon task, whether to trust its next action, abstain, or hand off to a human....
arXiv:2606. 00765v1 Announce Type: new Abstract: LLM-based agents increasingly solve complex tasks through long trajectories involving reasoning steps, tool calls, and inter-agent communication.
AgentPerturb introduces a systematic framework to evaluate large language model agents under controlled interaction perturbations, addressing the lack of unified evaluation methods for noisy real‑world tasks. The study quantifies robustness across frontier models, identifies task‑specific vulnerabilities, and shows that agents often fail to detect altered inputs, propagating errors and incurring higher computational costs. The authors provide a benchmark‑construction pipeline and plan to release code and data to support future robustness‑oriented training.
The paper introduces a method for monitoring web agents without relying on internal signals such as token logits. It proposes two observable trajectory representations—Macro features that capture cross‑step agent–environment interactions, and Micro features that assess consistency of intention, action, and expected state change via repeated black‑box queries. By labeling the first uncorrected critical error that leads to final failure as a key‑step boundary, the approach preserves valid early prefixes of failed trajectories and achieves risk prediction performance competitive with internal‑signal baselines across WebArena‑Lite and Online Mind2Web benchmarks.
Intent misinterpretation during vehicle interactions causes recurring planning failures. We study a decision layer in which a language-guided intent module reads structured descriptors, computes a smo...
RealControlBench is a new benchmark that evaluates fine‑grained steering of large language models and agents across diverse behavioral targets and interaction settings. It extends beyond isolated generation tasks to include complex, execution‑grounded, multi‑turn agent interactions, covering instruction‑following control, tool‑use decision control, reasoning performance, reasoning length, and safety awareness. The benchmark jointly measures target controllability and capability preservation, testing robustness across prompts, domains, and model architectures, and finds that current steering methods are often unreliable in realistic scenarios.
The paper examines whether internal representations of agentic systems can better indicate task success than traditional confidence measures. It introduces two methods—Latent Trajectory Dynamics (LTD) and Action Representation Probe (ARP)—that analyze changes in residual-stream representations and action-level representations, respectively. Experiments on Bash, SQL, and Python benchmarks with Qwen and DeepSeek models show these methods outperform conventional surface-level and sequence-based calibration baselines, offering a zero‑overhead reliability monitor without prompt changes or multiple rollouts.
arXiv:2605.27690v2 Announce Type: replace-cross Abstract: LLM agents increasingly operate through multi-turn tool use and environment interaction, where safety risks often emerge from intermediate st...
The paper introduces Confidence Reasoning Graphs (CRGs), an inference-time framework that estimates the probability an LLM agent has successfully completed its task using a single trajectory without requiring privileged model access or training data. CRGs decompose the agent’s overall claim of success into contextualized sub-claims, estimate confidence for each terminal claim based on trajectory evidence, and aggregate these into an overall confidence estimate. Experiments across multiple benchmarks, models, and agent frameworks show that CRGs produce better-calibrated confidence and more effective risk-aware decision making than existing verbalized, sampling-based, and white-box surrogate methods, while also providing transparent, auditable evidence for each estimate.