# VeriGrey: Greybox Agent Validation

arXiv: https://arxiv.org/abs/2603.17639 · HTML: https://mlsanity.com/paper/2603.17639/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Agent threats, guardrails and governance, one of 16 directions in LLM agents and tool use.
- Builds on 20 of its 43 references that are in the map.
- First posted on arXiv 18 Mar 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, 20 of 20)

- [Toolformer: Language Models Can Teach Themselves to Use Tools](https://mlsanity.com/paper/2302.04761/) (2023)
- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [SWE-bench: Can Language Models Resolve Real-World GitHub Issues?](https://mlsanity.com/paper/2310.06770/) (2023)
- [Competition-Level Code Generation with AlphaCode](https://mlsanity.com/paper/2203.07814/) (2022)
- [Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection](https://mlsanity.com/paper/2302.12173/) (2023)
- [Ignore Previous Prompt: Attack Techniques For Language Models](https://mlsanity.com/paper/2211.09527/) (2022)
- [A Survey on the Memory Mechanism of Large Language Model based Agents](https://mlsanity.com/paper/2404.13501/) (2024)
- [Prompt Injection attack against LLM-integrated Applications](https://mlsanity.com/paper/2306.05499/) (2023)
- [GPTFUZZER: Red Teaming Large Language Models with Auto-Generated Jailbreak Prompts](https://mlsanity.com/paper/2309.10253/) (2023)
- [AgentPoison: Red-teaming LLM Agents via Poisoning Memory or Knowledge Bases](https://mlsanity.com/paper/2407.12784/) (2024)
- [Defending Against Indirect Prompt Injection Attacks With Spotlighting](https://mlsanity.com/paper/2403.14720/) (2024)
- [Auto-GPT for Online Decision Making: Benchmarks and Additional Opinions](https://mlsanity.com/paper/2306.02224/) (2023)
- [The Landscape of Emerging AI Agent Architectures for Reasoning, Planning, and Tool Calling: A Survey](https://mlsanity.com/paper/2404.11584/) (2024)
- [Optimization-based Prompt Injection Attack to LLM-as-a-Judge](https://mlsanity.com/paper/2403.17710/) (2024)
- [IsolateGPT: An Execution Isolation Architecture for LLM-Based Agentic Systems](https://mlsanity.com/paper/2403.04960/) (2024)
- [AirGapAgent: Protecting Privacy-Conscious Conversational Agents](https://mlsanity.com/paper/2405.05175/) (2024)
- [AgentVigil: Generic Black-Box Red-teaming for Indirect Prompt Injection against LLM Agents](https://mlsanity.com/paper/2505.05849/) (2025)
- [Firewalls to Secure Dynamic LLM Agentic Networks](https://mlsanity.com/paper/2502.01822/) (2025)
- [GoEX: Perspectives and Designs Towards a Runtime for Autonomous LLM Applications](https://mlsanity.com/paper/2404.06921/) (2024)
- [PROMPTFUZZ: Harnessing Fuzzing Techniques for Robust Testing of Prompt Injection in LLMs](https://mlsanity.com/paper/2409.14729/) (2024)

## Built on by (most cited first, 1 of 1)

- [Know Your Agent: Reconnaissance-Driven Pentesting of AI Agents](https://mlsanity.com/paper/2607.19837/) (2026)

## Abstract (arXiv)

Agentic AI has been a topic of great interest recently. A Large Language Model (LLM) agent involves one or more LLMs in the back-end. In the front end, it conducts autonomous decision-making by combining the LLM outputs with results obtained by invoking several external tools. The autonomous interactions with the external environment introduce critical security risks. In this paper, we present a grey-box approach to explore diverse behaviors and uncover security risks in LLM agents. Our approach VeriGrey uses the sequence of tools invoked as a feedback function to drive the testing process. This helps uncover infrequent but dangerous tool invocations that cause unexpected agent behavior. As mutation operators in the testing process, we mutate prompts to design pernicious injection prompts. This is carefully accomplished by linking the task of the agent to an injection task, so that the injection task becomes a necessary step of completing the agent functionality. Comparing our approach with a black-box baseline on the well-known AgentDojo benchmark, VeriGrey achieves 33% additional efficacy in finding indirect prompt injection vulnerabilities with a GPT-4.1 back-end. We also conduct real-world case studies with the widely used coding agent Gemini CLI, and the well-known OpenClaw personal assistant. VeriGrey finds prompts inducing several attack scenarios that could not be identified by black-box approaches. In OpenClaw, by constructing a conversation agent which employs mutational fuzz testing as needed, VeriGrey is able to discover malicious skill variants from 10 malicious skills (with 10/10= 100% success rate on the Kimi-K2.5 LLM backend, and 9/10= 90% success rate on Opus 4.6 LLM backend). This demonstrates the value of a dynamic approach like VeriGrey to test agents, and to eventually lead to an agent assurance framework.