telemetry-analysis

Tag

Cards List
#telemetry-analysis

When Agentic Executions Fail: Detecting and Localizing Runtime Faults from Telemetry

arXiv cs.AI · 2026-08-18 Cached

The paper presents AgentChaosBench, a benchmark for detecting and localizing runtime faults in LLM-based agentic systems, and evaluates it using zero-shot LLM baselines, revealing significant challenges in fault diagnosis.

0 favorites 0 likes
← Back to home

Submit Feedback