Tool Use & Agents - Weekly Roundup

Edge service orchestration can achieve up to 64.5% lower decision latency than LLMs by using Jev decision models, even at high loads where LLMs fail.

The DeepSeek-V4.1-Flash model, a multimodal Mixture-of-Experts model with 552B backbone parameters and support for contexts of up to one million tokens, is introduced, substantially improving cost efficiency for agentic workloads and pushing the limits of KV cache compression.

RoboDawn is investigated, a human-intuitive interface that exposes robotic control to an agentic VLM through a compact set of discrete translation, rotation, and gripper commands, and an in-context learning (ICL) scheme is introduced that ground the VLM in both interface usage and task-solving strategies.

RoboDawn is investigated, a human-intuitive interface that exposes robotic control to an agentic VLM through a compact set of discrete translation, rotation, and gripper commands, and an in-context learning (ICL) scheme is introduced that ground the VLM in both interface usage and task-solving strategies.

Edge service orchestration can achieve up to 64.5% lower decision latency than LLMs by using Jev decision models, even at high loads where LLMs fail.

This work takes an RSI-inspired approach at the harness layer, scaling auto-research loops across increasingly numerous and diverse environments for harness rollouts, and yields reusable improvements that transfer beyond their development setting.

This paper proposes a conceptual Agentic AI architecture for AI systems that can plan, access data sources, execute tools, and act autonomously and audibly, aimed at supporting decision-making across the three Brazilian Armed Forces.

It is shown that a misaligned model can perform inference engine fingerprinting to determine the specific engine (e.g., vLLM, SGLang) which executes the model, and several ways that inference engines could be changed to make fingerprinting attacks more difficult are discussed.

GenAIMMD is introduced, an iterative algorithm that actively and self-consistently learns the ideal reaction coordinate (the committor) and trains a conditioned Boltzmann Generator to sample from arbitrary bias windows along it, providing a correlation-free and fully parallelizable path sampling scheme that does not require prior knowledge of the system's transition mechanism.

The semantic action graph is presented, a lightweight domain schema that represents a sports match as performer, action, recipient, moment, and state nodes connected by role, temporal, and outcome edges, providing early evidence that one small, human-readable schema can ground agent generation and support human interpretation at the same time.

This article investigates neuro-symbolic agentic AI (NSAAI) as a framework for combining neural grounding, symbolic reasoning, and closed-loop agentic interaction to support more reliable and adaptive UAV autonomy.

Inspired by the human annotation training process, ProKDA first uses an agentic background knowledge construction pipeline to obtain external knowledge related to meme understanding, and adopts a three-stage training strategy that sequentially performs background knowledge learning, hatefulness detection learning, and hatefulness boundary alignment.

The introduction of AutoData, an agent that searches directly over executable selection algorithms, suggests that data engineering can be treated as an agentic machine learning problem, extending autonomous research from model and training-code optimization to the data.

FARSIGHT (Financial Agent Robustness and Security Investigation and Global Holistic Testing), a framework that performs scheme-level evaluation of financial LLM agents on two axes: robustness under market turbulence and security against three attack types: attacks on information sources, attacks on agents, and agent-as-attacker behaviors.

VideoResearcher achieves state-of-the-art performance among self-improving agents and approaches the human-designed upper bound, demonstrating a training-free paradigm for long-video understanding that expands agent capabilities through autonomous tool development while reducing costly manual engineering.

A unified four-step post-training pipeline comprising a learned temporal hand-action codec, supervised fine-tuning, DAgger, and real-world residual reinforcement learning provides a practical path for adapting VLA foundation models to reliable real-world dexterous manipulation.

This work proposes StageGuard, an agentic distillation framework for accurate and efficient stage-transition decisions, which combines teacher-model reasoning with demonstration trajectories to generate structured explanations of subtask completion and policy switching.

Results show that the multi-agent design of RoboFind fits the demands of personalized object search, where verifying object identity before declaring completion is what makes the outcome something a user can rely on.

SELF-INDEX is proposed, a framework that enables an index to self-evolve without human intervention, and its Optimizer autonomously diagnoses retrieval shortfalls, selectively revises the responsible index keys, and validates each revision before updating the index.

By centering smart glasses as wearable-intelligence platforms, this survey provides a unified framework for organizing technologies, applications, and open challenges in this emerging area.

An autonomous AI agent integrating Reasoning and Acting with Retrieval-Augmented Generation with Retrieval-Augmented Generation is developed to classify 10,211 Human Phenotype Ontology terms, enabling standardized panel design by providing reliable, automated classification supported by direct evidence.

These results support extending coefficient conditioning with explicit equation relationships for physics-based solver adaptation, and support extending coefficient conditioning with explicit equation relationships for physics-based solver adaptation.

The proposed scalable trust discovery architecture incorporates a registry-suffix-anchored composite identity scheme, which binds an agent native identifier to a trusted registry suffix to generate a globally discoverable identity.

UnifiedPlayers, a cooperative framework comprising a Planning Player that generates tasks, an Execution Player that produces multi-turn trajectories with Python tool calls, and an Evaluation Player that constructs executable verifiers, highlights cooperation among specialized players as a promising path toward self-enhanced tool-integrated agents.

The latency with respect to the resource dynamics of processing multiple requests and tasks concurrently is characterized and new optimization opportunities that exploit the resource dynamics of tasks are demonstrated: CPU-aware tool admission and task-aware CPU allocation.

The findings of MaSCoD support structural pre-organization as an explicit design and evaluation target for omission control and motivate evaluating context construction jointly with its utilization in judgment.

These findings support long-context inference in which pretrained models guide their own access to the information they retain, by introducing On-Demand Attention (ODA), a local-first decoding method that uses a lightweight recall head to selectively invoke global attention as its predicted benefit changes during generation.

The results show that reliable scenario generation requires verifying realized behavior, not merely executable code, and demonstrate the value of explicit intermediate representations for verifiable and repairable language-driven co-simulation.

The goal of PrefixBench-H100 is to expose the practical operating envelope of prefix reuse for H100-class LLM serving, and identifies the regime where prefix reuse provides substantial first-token latency reductions and the regime where cache pressure erodes them.

The DeepSeek-V4.1-Flash model, a multimodal Mixture-of-Experts model with 552B backbone parameters and support for contexts of up to one million tokens, is introduced, substantially improving cost efficiency for agentic workloads and pushing the limits of KV cache compression.

This work proposes RetireOPD (Self-Retiring On-Policy Distillation), which first optimizes a decoupled, skill-conditioned teacher with environment rewards and then trains a skill-free student jointly with RL and OPD.

This work proposes DualSQL, a new Text-to-SQL system consisting of two agents powered by a single model backbone, enabling joint optimization through a robust multi-agent reinforcement learning (RL) framework, and introduces a new SQL correctness metric, robust execution match (REX), to more accurately judge SQL correctness and assign reward signals.

WAVE-Go, an image-goal navigation framework that separates world-action prediction from interruptible command execution, and its executor adaptively selects an action prefix and cancels pending commands when updated observations invalidate execution, support adaptive, interruptible execution as a means of balancing navigation performance and planning overhead.

Finder is presented, an agentic closed-loop object-finding primitive for embodied grounding that maintains a typed loop state that links query-conditioned planning, scoped evidence gathering, candidate verification, and accept/continue/abort control.

This work instantiates Real-Time EXPO-FT, an RL framework for finetuning real-time VLA policies that meets the real-time control requirements of dynamic real-world manipulation, demonstrating rapid, sample-efficient adaptation to challenging real-world dynamics.

This work introduces PACT (Pressure-Applied Compliance Testing), a benchmark for rule-following under pressure in AI agents assisting employees in daily tasks across twelve regulated enterprise domains and forty-eight scenarios, each set in a realistic multi-turn conversation.

This work proposes that tool calls report their progress explicitly while they run, and measures what that takes, and shows that at the points where a KV cache decision is made, the reported progress is between several times and an order of magnitude more accurate than the best published predictors, and it stays accurate when the environment changes.

The results identify information required for recovery selection, establish conditional guarantees against harmful updates, and quantify the recovery improvements forgone through conservative acceptance.

This project verified that generating robust and reliable software using agentic AI requires careful practice and human supervision and demonstrated that using agentic AI workflow for software development (with its prompts and skills) can become a specification of the code that leads to cost-effective sustainability of the software.

The model is given a formal basis by transplanting the beta-factor model of common-cause failure from reliability engineering, a seven-step protocol whose outputs a third party can verify, a structural detectability analysis of a procurement-controls agent audited at three grades, and a Monte Carlo study of the model.

A time-aware assessment framework that models avoided emissions and AI-induced emissions as discrete-time streams over a finite time horizon is presented and provides decision guidelines that support go/no-go screening, timing decisions, and minimum"bang-for-your-buck"thresholds.

A framework to measure the trust profile of an arbitrary LLM based on a variety of payload framings sent through different channels is presented, and cross-channel fragmentation attacks that distribute seemingly benign payloads across two or three channels are devised.

This study shows that without explicit network-layer indication, MCP traffic structurally and temporally resembles Command and Control beaconing behavior, specifically the polling architectures used by advanced persistent threats like Cobalt Strike, and proposes an agent-native network indication standard including Agent-Native ALPN and standardized out-of-band headers.

Conditioning on outcome class shows that response reliability and correctness are inversely related, whereas genuinely computing methods answered least often and were correct 19.3%.

It is argued that harnessing physical-digital AI agents, through persistent coupling of digital intelligence with physical sensing, dynamics, action, and feedback, provides a system-level pathway toward adaptive and continuously evolving UAV autonomy.

This work presents OmniRisk, an omnidirectional planning framework that learns trajectory-level risk offline for efficient onboard evasion and enables efficient onboard avoidance, with real-world flights demonstrating consecutive evasive maneuvers at relative encounter speeds up to 15 m/s without fine-tuning.

GPT-Policy is introduced, a general-agent framework for in-context robot learning that integrates a context compiler that preserves task-relevant visual transitions, a VLM that proposes robot-tool actions, and a constrained controller that verifies and executes each action and reports its outcome.

This work proposes DualViewEval, an agent benchmark compression method that jointly exploits outcome and process relations to learn an exact-size miniset and predict the full-benchmark scores, and achieves the best results in all datasets.

Backend delegation is an effective and modular approach for combining natural duplex speech interaction with strong agentic tool-call capabilities and significantly outperforms GPT-realtime-mini and Qwen3-Omni-30B-A3B-Instruct on EVA-Bench.

The paper distinguishes defined competence from demonstrated capability and positions IDDL as an interoperable behavioural assessment and certification layer that can operate downstream of international competency frameworks.

This work introduces a unified multi-agent framework, MAGS, that generates executable programs with formal safety guarantees, using Dafny as a verification-aware intermediate representation where safety properties can be mechanically checked.

It is proved hallucination defense must precede any causal gate, and characterize the one irreducible residue (borrowed arguments schema-indistinguishable from a valid call), and the versioned Hallucinated-Tools Benchmark is released so any resolver is comparable across submissions.

A three-tier observability framework comprising catalog metadata, passive compliance signals, and live vulnerability analysis is applied to empirically characterize the public MCP server ecosystem, identifying a Security-Observability Tradeoff observed in the current ecosystem.

A causal-enhanced approach orchestrated end-to-end by an Agentic Controller with self-reflective closed-loop control with self-reflective closed-loop control is proposed, which includes two core modules to address the above two challenges.

PentestChain is presented, a ten-phase automated penetration testing framework that couples a curated, deterministic exploit map with a cost-aware AI cascade-a local Ollama model, kept off the critical path by a deterministic backbone, sustains end-to-end operation at zero measured paid-API cost.

Results show that grounding approval in the action that will actually execute provides architectural protection against security failures that prompt-level defences and conventional HITL confirmation cannot reliably prevent.

This work presents Cairn, a community reputation platform that captures collective knowledge, allowing agents to query the community's opinion of a resource before use and to submit evidence-backed ratings afterward, and evaluates Cairn's reputation engine under adversarial simulation.

This article categorizes existing UAV inspection architectures, identifies their key system challenges and architectural requirements, and experimentally assesses the feasibility of semantic edge intelligence on NVIDIA Jetson UAV-class hardware using the COCO-Bridge dataset.

AutoTuring hands the same agent the same 15-dimensional accelerator space twice: once as named architectural knobs with simulator counters, once as anonymous variables on [0,1], with the evaluator, the legal space and the reachable optima held identical, so that the only thing that varies is whether the problem means anything.

The results show that agentic search can support density-functional development when flexible hypothesis generation is coupled to automated physical verification, and show that agentic search can support density-functional development when flexible hypothesis generation is coupled to automated physical verification.

This work demonstrates contrastive scoring, candidate-pool reranking, and weighted ranking as composable queries as composable queries, alongside executable SQL and PyTerrier counterparts.

This first study of Web search across four major conversational platforms is presented, combining real-world user interactions with controlled experiments using the same platform's models by their APIs (invitro), finding that Web-search decisions vary substantially across platforms and models.

An initial benchmark dataset is introduced, namely ASURRE, for AI-assisted survey participation to capture usage strategies ranging from full generation and revision to persona-grounded agentic completion, and it is shown that agentic completion cannot fully replicate respondent-level behaviour and leaves distinctive behavioural traces.

EchoPath is introduced, a model-agnostic harness that converts artifact-validated GUI trajectories into standardized, parameter-controlled callable memories, analogous to Model Context Protocol (MCP)-style tool calls rather than unstructured experience records, which support a bounded form of enterprise GUI memory.

Experiments show that EvolveTrade often improves Sharpe Ratio and Cumulative Return over fixed-policy LLM baselines, achieving the improved SR and CR in most evaluated settings, and suggest that adapting the reusable procedure governing tool use is a key direction for building more robust LLM trading agents.

The results suggest that model-level alignment is not compositional: individually capable and apparently safe agents can form systems with qualitatively different failure modes as AI becomes persistent and interconnected.

A framework for evaluating data-science agents on live, continuously updated data using executable ground truth and format-agnostic factoid scoring, which achieves a 29% improvement in the Matthews Correlation Coefficient and a 16% reduction in token consumption per test case.

AURA (Agentic Understanding and Refinement of recommender Algorithms), an end-to-end agentic system that performs qualitative evaluation at scale and can then generate improvements to the authors' algorithms at the code level, is presented.

The Robot Data Factory is introduced, a mission-driven infrastructure and methodology for continuously generating, validating, benchmarking, and reusing robot experience that reframes robot data generation as a continuous scientific production process and provides a pathway toward reproducible, scalable, and eventually federated infrastructure for Physical AI.

What the boom left behind of the OpenClaw AI agent went viral, and its public skill registry boomed is measured, drawing on the OpenClaw Git history, its GitHub issues and pull requests, and three ClawHub registry snapshots.

This work model agents as both arbitrage extractors and Maximal Extractable Value targets, derive the optimal trade size for a risk-averse agent under mean-variance utility with stochastic bridge delays, and formalize multi-chain path selection as a belief-weighted online learning problem whose belief estimates converge under a Robbins-Monro schedule.

An automated framework is developed that constructs user runtime instruction-agnostic UI desynchronization attacks and realizes them in deployable APKs, demonstrating that this threat is feasible and feasible.

A comprehensive security analysis on autonomous AI-penetration testing agents is carried out, which systematically analyse representative agent architectures, characterise their trust boundaries and attack surfaces and proposes a threat taxonomy that is aligned with the lifecycle and covers LLM lifecycle attacks, agent-architecture attacks and cross-cutting behavioural attacks.

This work presents an agentic, LLM-based framework that automatically detects potential log file exposures in plugins of the most popular CMS (WordPress), and derives a taxonomy of log file path and protection patterns and deduce a set of best practices for developers to securely handle them.