Tracing Agentic Failure from the Flow of Success
Failure attribution for LLM-based agentic systems, i.e., identifying which steps in a failure trajectory caused the task to fail, is critical for debugging and improving these systems. Existing approaches either rely on prompting-based pipelines, which are computationally expensive, or require post-training on failure trajectories with step-level error annotations, which are costly to collect and difficult to scale. We argue that a practical failure attribution model should be lightweight and trainable without step-level supervision on failure data. To this end, we address unsupervised failure
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- PossiblePossibly related (embedding) · 56%AgentTrace →
- PossiblePossibly related (embedding) · 54%A system-level approach to prompt injection: separating instruction and data channels in LLM agents [P] →
- PossiblePossibly related (embedding) · 51%Launch HN: Agnost AI (YC S26) – Extract user feedback from agent conversations →
- PossiblePossibly related (embedding) · 51%Prompt injection is exploiting enterprise AI's biggest design flaws by targeting agents, RAG pipelines and model routers →
- PossiblePossibly related (embedding) · 50%agent-tools →
- LinkedLinked via arxiv author · 85%Samuel Yeh →
“Tracing Agentic Failure from the Flow of Success”
- LinkedLinked via arxiv author · 85%Yiwen Zhu →
“Tracing Agentic Failure from the Flow of Success”
- LinkedLinked via arxiv author · 85%Shaleen Deep →
“Tracing Agentic Failure from the Flow of Success”