aiqbench analysis
MetaPI addresses the critical gap in security evaluation for LLM agents by converting existing utility benchmarks into comprehensive indirect prompt injection (IPI) test suites. The toolkit automates the generation of adaptive attack payloads and validates defenses within realistic, high-privilege environments, revealing that current solutions struggle to balance strong utility with robust security. This work is essential for advancing the safety and reliability of autonomous agents operating in complex digital ecosystems.
MetaPI: Constructing Prompt Injection Benchmarks from LLM Agent Benchmark
From arXiv
Full textarXiv:2611.10026v1 Announce Type: cross Abstract: Large language model (LLM) agents increasingly perform complex tasks using high-privilege tools, making indirect prompt injection (IPI) a critical security risk. However, IPI benchmarks lag behind their rapidly evolving utility counterparts, only offering limited task and environment coverage and relying on narrow, static attack scenarios. In this paper, we propose meta-benchmarking for IPI: transforming existing utility benchmarks into end-to-end IPI evaluation suites that reuse their realistic tasks, environments, and infrastructure.
We present MetaPI, an extensible toolkit that automatically identifies plausible injection points, generates realistic attack objectives, optimizes adaptive payloads, and validates attacks in the real environment. Its modular interfaces support heterogeneous environments, agents, defenses, and attack algorithms, allowing new components to be integrated without redesigning the evaluation pipeline. We apply MetaPI to AppWorld, VitaBench, and Terminal-Bench and evaluate representative agents and defenses across multiple backbone LLMs. Our results show that no evaluated defense simultaneously achieves strong utility and strong security in realistic, adaptive agentic IPI settings. MetaPI thus enables scalable and consistent security evaluation across diverse and evolving agent ecosystems
Abstract:Large language model (LLM) agents increasingly perform complex tasks using high-privilege tools, making indirect prompt injection (IPI) a critical security risk. However, IPI benchmarks lag behind their rapidly evolving utility counterparts, only offering limited task and environment coverage and relying on narrow, static attack scenarios. In this paper, we propose meta-benchmarking for IPI: transforming existing utility benchmarks into end-to-end IPI evaluation suites that reuse their realistic tasks, environments, and infrastructure.
We present MetaPI, an extensible toolkit that automatically identifies plausible injection points, generates realistic attack objectives, optimizes adaptive payloads, and validates attacks in the real environment. Its modular interfaces support heterogeneous environments, agents, defenses, and attack algorithms, allowing new components to be integrated without redesigning the evaluation pipeline. We apply MetaPI to AppWorld, VitaBench, and Terminal-Bench and evaluate representative agents and defenses across multiple backbone LLMs. Our results show that no evaluated defense simultaneously achieves strong utility and strong security in realistic, adaptive agentic IPI settings. MetaPI thus enables scalable and consistent security evaluation across diverse and evolving agent ecosystems
Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
arXiv:2611.10026 [cs.CR]
(or arXiv:2611.10026v1 [cs.CR] for this version)
https://doi.org/10.48550/arXiv.2611.10026 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 268 words
Related coverage
Embodied-BenchClaw: An Autonomous Multi-Agent System for Embodied Spatial Intelligence Benchmark Construction
arXiv · Jun 11, 2026
Embodied-BenchForge: A Closed-Loop Agentic Workflow for Embodied Benchmark Construction
arXiv · Sep 14, 2026
MedBenchAgent: Towards Systematic Automation of Medical VLM Benchmark Construction
arXiv · Oct 10, 2026
Intelligent Automation for Embodied Benchmark Construction: Pipelines, Embodiments, Simulators, and Trends
arXiv · Jun 11, 2026