…Measuring Runtime Trust Failures in Third-Party Agent Skills (2026) AgentVisor: Defending LLM Agents Against Prompt Injection via Semantic Virtualization (2026) ClawGuard: A Runtime Security Framework for Tool-Augmented LLM Agents Against…
…We observed improvements in Agent's environment interaction KPI such as tool success rate and tool parameter accuracy . The framework also offers new metrics such as Execution Quality to generate new insights…
…To answer it, we introduce Pi-Serini (= PI + Anserini), a minimal BM25-based search agent equipped with search , browse , and read tools. This interface allows agents to cache retrieved rankings locally and…
…Benchmarking Spatial-Functional Intelligence in Multimodal LLMs (2026) Affordance Agent Harness: Verification-Gated Skill Orchestration (2026) VTC-Bench: Evaluating Agentic Multimodal Models via Compositional Visual Tool Chaining (2026) A3R: Agentic Affordance Reasoning…
…Towards Training a Multimodal Agentic Search Model from Scratch (2026) E3-TIR: Enhanced Experience Exploitation for Tool-Integrated Reasoning (2026) PRAISE: Prefix-Based Rollout Reuse in Agentic Search Training (2026) $S^3…
…Flexibility, Cost, and Bottlenecks in Multi-Agent Workflows (2026) ClawForge: Generating Executable Interactive Benchmarks for Command-Line Agents (2026) Utility-Guided Agent Orchestration for Efficient LLM Tool Use (2026) AgentForge: Execution-Grounded…
…Encouraging Tool Exploration for Deep Research with Synthetic Plans (2026) GraphWalker: Agentic Knowledge Graph Question Answering via Synthetic Trajectory Curriculum (2026) LiteResearcher: A Scalable Agentic RL Training Framework for Deep Research Agent…
…AI-generated summary LLM-based autonomous agents have demonstrated strong capabilities in reasoning, planning, and tool use, yet remain limited when tasks require sustained coordination across roles, tools, and environments. Multi-agent…
…An Agentic Benchmark for Novel API Acquisition (2026) RaTA-Tool: Retrieval-based Tool Selection with Multimodal Large Language Models (2026) FitText: Evolving Agent Tool Ecologies via Memetic Retrieval (2026) ToolOmni: Enabling Open…
…We present a comprehensive empirical study of multi-turn agentic RL for medical AI, built on , a gymnasium-compatible environment spanning 10 clinical domains with 3.6K+ tasks, 135 domain-specific tools…