SNARE: Adaptive Scenario Synthesis for Eliciting Overeager Behavior in Coding Agents

Abstract

A coding agent executes a benign task as a sequence of shell, file, and network actions, any of which can quietly exceed the authorized scope while the task still completes. We call this overeager behavior: the prompt is not adversarial and the run succeeds, yet an out-of-scope step can leak credentials or delete files. Existing benchmarks miss it: task-completion suites credit any finished run, jailbreak suites probe adversarial prompts, and the one prior overeager benchmark applies a single fi

Categories

Framework mappings

Suggested from the entry's categories.

@misc{qu2026snare,
  title = {{SNARE: Adaptive Scenario Synthesis for Eliciting Overeager Behavior in Coding Agents}},
  author = {Yubin Qu and Yi Liu and Gelei Deng and Yanjun Zhang and Yuekang Li and Ying Zhang and Leo Yu Zhang},
  year = {2026},
  month = may,
  eprint = {2605.28122},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2605.28122}
}