# Toward a Principled Framework for Agent Safety Measurement

arXiv: https://arxiv.org/abs/2605.01644 · HTML: https://mlsanity.com/paper/2605.01644/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Agent threats, guardrails and governance, one of 16 directions in LLM agents and tool use.
- Builds on 15 of its 22 references that are in the map.
- First posted on arXiv 2 May 2026.

## Placed in

- LLM agents and tool use / [Agent threats, guardrails and governance](https://mlsanity.com/agents/agent-threats-guardrails-and-governance/)

## Builds on (most cited first, 15 of 15)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [The Curious Case of Neural Text Degeneration](https://mlsanity.com/paper/1904.09751/) (2019)
- [Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection](https://mlsanity.com/paper/2302.12173/) (2023)
- [The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructions](https://mlsanity.com/paper/2404.13208/) (2024)
- [AgentHarm: A Benchmark for Measuring Harmfulness of LLM Agents](https://mlsanity.com/paper/2410.09024/) (2024)
- [AgentPoison: Red-teaming LLM Agents via Poisoning Memory or Knowledge Bases](https://mlsanity.com/paper/2407.12784/) (2024)
- [Agent Security Bench (ASB): Formalizing and Benchmarking Attacks and Defenses in LLM-based Agents](https://mlsanity.com/paper/2410.02644/) (2024)
- [R-Judge: Benchmarking Safety Risk Awareness for LLM Agents](https://mlsanity.com/paper/2401.10019/) (2024)
- [Defending Against Indirect Prompt Injection Attacks With Spotlighting](https://mlsanity.com/paper/2403.14720/) (2024)
- [Constitutional Classifiers: Defending against Universal Jailbreaks across Thousands of Hours of Red Teaming](https://mlsanity.com/paper/2501.18837/) (2025)
- [A Thorough Examination of Decoding Methods in the Era of LLMs](https://mlsanity.com/paper/2402.06925/) (2024)
- [ToolSword: Unveiling Safety Issues of Large Language Models in Tool Learning Across Three Stages](https://mlsanity.com/paper/2402.10753/) (2024)
- [BrowseSafe: Understanding and Preventing Prompt Injection Within AI Browser Agents](https://mlsanity.com/paper/2511.20597/) (2025)
- [Toward Principled LLM Safety Testing: Solving the Jailbreak Oracle Problem](https://mlsanity.com/paper/2506.17299/) (2025)

## Abstract (arXiv)

LLM agents emit actions, not just text, and once taken, those actions often cannot be undone. Yet today's agent-safety evaluations run greedy or a few sampled rollouts and report a single safe/unsafe rate -- blind to the long-tail trajectories where unsafe behavior may arise from low-probability but non-negligible actions. We argue agent safety should be measured by search, not sampling. We apply BOA, a framework that, given a deployment configuration (model, decoder, prompt, environment, judger, likelihood budget), searches the in-budget trajectory space and reports a safety score: the probability the agent stays safe under the configuration. BOA searches both within a single LLM round and across the agent-environment interaction tree under a given likelihood budget, and makes search practical via batched decoding/judging, prefix caching, and chunked tree expansion. On agent-safety workloads, BOA discovers unsafe trajectories that greedy and sampled evaluations miss. BOA can additionally be used for ranking models, defenses, and attacks, all on the same scale, with manageable GPU costs.