Near-Miss: Latent Policy Failure Detection in Agentic Workflows
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909042774900736 |
|---|---|
| author | Rabinovich, Ella Boaz, David Zwerdling, Naama Anaby-Tavor, Ateret |
| author_facet | Rabinovich, Ella Boaz, David Zwerdling, Naama Anaby-Tavor, Ateret |
| contents | Agentic systems for business process automation often require compliance with policies governing conditional updates to the system state. Evaluation of policy adherence in LLM-based agentic workflows is typically performed by comparing the final system state against a predefined ground truth. While this approach detects explicit policy violations, it may overlook a more subtle class of issues in which agents bypass required policy checks, yet reach a correct outcome due to favorable circumstances. We refer to such cases as near-misses or latent failures. In this work, we introduce a novel metric for detecting latent policy failures in agent conversations traces. Building on the ToolGuard framework, which converts natural-language policies into executable guard code, our method analyzes agent trajectories to determine whether agent's tool-calling decisions where sufficiently informed. We evaluate our approach on the $τ^2$-verified Airlines benchmark across several contemporary open and proprietary LLMs acting as agents. Our results show that latent failures occur in 8-17% of trajectories involving mutating tool calls, even when the final outcome matches the expected ground-truth state. These findings reveal a blind spot in current evaluation methodologies and highlight the need for metrics that assess not only final outcomes but also the decision process leading to them. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_29665 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Near-Miss: Latent Policy Failure Detection in Agentic Workflows Rabinovich, Ella Boaz, David Zwerdling, Naama Anaby-Tavor, Ateret Computation and Language Agentic systems for business process automation often require compliance with policies governing conditional updates to the system state. Evaluation of policy adherence in LLM-based agentic workflows is typically performed by comparing the final system state against a predefined ground truth. While this approach detects explicit policy violations, it may overlook a more subtle class of issues in which agents bypass required policy checks, yet reach a correct outcome due to favorable circumstances. We refer to such cases as near-misses or latent failures. In this work, we introduce a novel metric for detecting latent policy failures in agent conversations traces. Building on the ToolGuard framework, which converts natural-language policies into executable guard code, our method analyzes agent trajectories to determine whether agent's tool-calling decisions where sufficiently informed. We evaluate our approach on the $τ^2$-verified Airlines benchmark across several contemporary open and proprietary LLMs acting as agents. Our results show that latent failures occur in 8-17% of trajectories involving mutating tool calls, even when the final outcome matches the expected ground-truth state. These findings reveal a blind spot in current evaluation methodologies and highlight the need for metrics that assess not only final outcomes but also the decision process leading to them. |
| title | Near-Miss: Latent Policy Failure Detection in Agentic Workflows |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2603.29665 |