paperAugust 2026Unreviewed

ECLIPSE: Self-Evolving Stealthy Prompt Injection Attack against Long-Horizon Agentic Systems

Shiqian Zhao, Yangfan Zhou, Xinfeng Li, Runyi Hu, Yechao Zhang, Yi Xie, Tianwei Zhang, Luu Anh Tuan

Abstract

Recently, large language model (LLM) agents, such as Codex, Claude Code, and OpenClaw, have become capable of planning and executing long-horizon tasks through repeated tool calls. This capability also creates new opportunities for prompt injection. Existing attacks either place the malicious objective in one explicit instruction, making it easy to detect, or distribute the intent across multiple execution stages, making successful completion unreliable. In this work, we propose ECLIPSE, a self-

Framework mappings

OWASP Top 10 for LLM Applications

LLM01Prompt Injection

MITRE ATLAS

AML.T0051LLM Prompt Injection

Suggested from the entry's categories.

Cite

@misc{zhao2026eclipse,
  title = {{ECLIPSE: Self-Evolving Stealthy Prompt Injection Attack against Long-Horizon Agentic Systems}},
  author = {Shiqian Zhao and Yangfan Zhou and Xinfeng Li and Runyi Hu and Yechao Zhang and Yi Xie and Tianwei Zhang and Luu Anh Tuan},
  year = {2026},
  month = aug,
  eprint = {2608.30441},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2608.30441}
}