CatchBench: When Can an Agent Failure Be Caught?
cs.LG, cs.MA, cs.PF
Submitted: 2026-08-24
Updated: 2026-09-17
Comments: 55 pages, 15 figures, 27 tables. Updated author list, analyses, and presentation. Code and data: https://github.com/yzhao062/catchbench
Code: https://github.com/yzhao062/catchbench
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- AgentRx: Diagnosing AI Agent Failures from Execution Trajectories
- TRAIL: Trace Reasoning and Agentic Issue Localization
- SentinelAgent: Graph-based Anomaly Detection in Multi-Agent Systems
- FORTIS: Benchmarking Over-Privilege in Agent Skills
- Who&When Pro: Can LLMs Really Attribute Failures in AI Agents?
- GAMMAF: A Common Framework for Graph-Based Anomaly Monitoring Benchmarking in LLM Multi-Agent Systems
- Formal Policy Enforcement for Agentic Systems
- Do Coding Agents Understand Least-Privilege Authorization?
- When Lower Privileges Suffice: Investigating Over-Privileged Tool Selection in LLM Agents
- AgenTracer: Who Is Inducing Failure in the LLM Agentic Systems?
- GRADE: Graph Representation of LLM Agent Dependency and Execution
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks