AI agents that leverage Large Language Models (LLMs) are increasingly becoming core building blocks of modern software systems. A wide range of frameworks is now available to support the specification of such applications. These frameworks enable the definition of agent setups using natural language prompting, which specifies the roles, goals, and tools assigned to the various agents involved. Within such setups, agent behavior is non-deterministic for any given input, highlighting the critical need for robust debugging and observability tools. In this work, we explore the use of process and causal discovery applied to agent execution trajectories as a means of enhancing developer observability. This approach aids in monitoring and understanding the emergent variability in agent behavior. Additionally, we complement this with LLM-based static analysis techniques to distinguish between intended and unintended behavioral variability. We argue that such instrumentation is essential for giving developers greater control over evolving specifications and for identifying aspects of functionality that may require more precise and explicit definitions.

The ability of LLM agents to plan and invoke tools exposes them to new safety risks, making a comprehensive red-teaming system crucial for discovering vulnerabilities and ensuring their safe deployment. We present SIRAJ: a generic red-teaming framework for arbitrary black-box LLM agents. We employ a dynamic two-step process that starts with an agent definition and generates diverse seed test cases that cover various risk outcomes, tool-use trajectories, and risk sources. Then, it iteratively constructs and refines model-based adversarial attacks based on the execution trajectories of former attempts. To optimize the red-teaming cost, we present a model distillation approach that leverages structured forms of a teacher model's reasoning to train smaller models that are equally effective. Across diverse evaluation agent settings, our seed test case generation approach yields 2 -- 2.5x boost to the coverage of risk outcomes and tool-calling trajectories. Our distilled 8B red-teamer model improves attack success rate by 100%, surpassing the 671B Deepseek-R1 model. Our ablations and analyses validate the effectiveness of the iterative framework, structured reasoning, and the generalization of our red-teamer models.

Recent advances in large language models (LLMs) and agent system designs have empowered agents with unprecedented levels of capability. However, existing agent benchmarks are showing a trend of rapid ceiling-hitting by newly developed agents, making it difficult to meet the demands for evaluating agent abilities. To address this problem, we propose the Trajectory-based V alidated-by-Reproducing Agent-benchmark Complexity Evolution (TRACE) framework. This framework takes an original task from an existing benchmark and encourages agents to freely explore and evolve it into a new task with higher difficulty while recording validatable agent trajectories. The framework proceeds in three stages: (1) evolutionary proposal mining, which provides task evolution proposals through preliminary exploration and divergent thinking; (2) problem formation and free exploration, where proposals are conceptualized into feasible problem candidates and the agents then explore them freely while recording their execution trajectories; and (3) multi-level validation, which ensures that the evolved tasks are accompanied by validatable and reproducible trajectories. Experiments on the GAIA benchmark demonstrate that the TRACE framework consistently enhances task complexity while improving the reliability of correctness through validatable execution trajectories. In addition, our framework can successfully adapt to and improve reasoning datasets represented by AIME-2024. This work marks a paradigm shift from static, manually curated benchmarks to dynamic, self-evolving evaluation systems, providing a sustainable and challenging runway for agent development.

Large Language Model Powered Multi-Agent Systems (MASs) are increasingly employed to automate complex real-world problems, such as programming and scientific discovery. Despite their promising, MASs are not without their flaws. However, failure attribution in MASs - pinpointing the specific agent actions responsible for failures - remains underexplored and labor-intensive, posing significant challenges for debugging and system improvement. To bridge this gap, we propose FAMAS, the first spectrum-based failure attribution approach for MASs, which operates through systematic trajectory replay and abstraction, followed by spectrum analysis.The core idea of FAMAS is to estimate, from variations across repeated MAS executions, the likelihood that each agent action is responsible for the failure. In particular, we propose a novel suspiciousness formula tailored to MASs, which integrates two key factor groups, namely the agent behavior group and the action behavior group, to account for the agent activation patterns and the action activation patterns within the execution trajectories of MASs. Through expensive evaluations against 12 baselines on the Who and When benchmark, FAMAS demonstrates superior performance by outperforming all the methods in comparison.

Large language models (LLMs) have shown remarkable capabilities in solving complex tasks. Recent work has explored decomposing such tasks into subtasks with independent contexts. However, some contextually related subtasks may encounter information loss during execution, leading to redundant operations or execution failures. To address this issue, we propose a training-free framework with an interaction mechanism, which enables a subtask to query specific information or trigger certain actions in completed subtasks by sending requests. To implement interaction, we introduce a subtask trajectory memory to enable resumption of completed subtasks upon receiving interaction requests. Additionally, we propose a new action during execution, which generates a concise and precise description of execution process and outcomes of a subtask, to assist subsequent subtasks in determining interaction targets and requests. We evaluate our framework on interactive decision-making task WebShop and multi-hop question answering HotpotQA, with GPT -3.5 and GPT -4, and comparison results show that our framework outperforms the state-of-the-art training-free baselines.

This paper introduces Investigate-Consolidate-Exploit (ICE), a novel strategy for enhancing the adaptability and flexibility of AI agents through inter-task self-evolution. Unlike existing methods focused on intra-task learning, ICE promotes the transfer of knowledge between tasks for genuine self-evolution, similar to human experience learning. The strategy dynamically investigates planning and execution trajectories, consolidates them into simplified workflows and pipelines, and exploits them for improved task execution. Our experiments on the XAgent framework demonstrate ICE's effectiveness, reducing API calls by as much as 80% and significantly decreasing the demand for the model's capability. Specifically, when combined with GPT-3.5, ICE's performance matches that of raw GPT-4 across various agent tasks. We argue that this self-evolution approach represents a paradigm shift in agent design, contributing to a more robust AI community and ecosystem, and moving a step closer to full autonomy.

We observe a large variety of robots in terms of their bodies, sensors, and actuators. Given the commonalities in the skill sets, teaching each skill to each different robot independently is inefficient and not scalable when the large variety in the robotic landscape is considered. If we can learn the correspondences between the sensorimotor spaces of different robots, we can expect a skill that is learned in one robot can be more directly and easily transferred to other robots. In this paper, we propose a method to learn correspondences among two or more robots that may have different morphologies. To be specific, besides robots with similar morphologies with different degrees of freedom, we show that a fixed-based manipulator robot with joint control and a differential drive mobile robot can be addressed within the proposed framework. To set up the correspondence among the robots considered, an initial base task is demonstrated to the robots to achieve the same goal. Then, a common latent representation is learned along with the individual robot policies for achieving the goal. After the initial learning stage, the observation of a new task execution by one robot becomes sufficient to generate a latent space representation pertaining to the other robots to achieve the same task. We verified our system in a set of experiments where the correspondence between robots is learned (1) when the robots need to follow the same paths to achieve the same task, (2) when the robots need to follow different trajectories to achieve the same task, and (3) when complexities of the required sensorimotor trajectories are different for the robots. We also provide a proof-of-the-concept realization of correspondence learning between a real manipulator robot and a simulated mobile robot.