Tag
提出了一种名为 Synthesis Through Simulation (STS) 的无模式数据合成范式，通过让 LLM 智能体在模拟企业环境中调用策略强制 API 来生成结构上天然有效的数据，其 Generalist Populator 智能体在十个环境中达到 0.88 的平均分布保真度和 100% 的约束满足率，且无需访问数据库 schema。
EdgeGen is a synthetic task generation framework that creates database-grounded edge-case tasks to improve tool-calling agents through fine-tuning and harness optimization, demonstrating consistent performance improvements.
This paper introduces VATS, a mutation-driven framework that systematically evolves adversarial payloads to exploit error-path injection in MCP-based tool-calling agents. It demonstrates that error messages with implicit authority triple the success rate of standard indirect prompt injection across frontier models.
SynAE is a framework for evaluating the quality of synthetic data used in tool-calling agent evaluations, assessing validity, fidelity, and diversity across multiple axes. It addresses challenges of insufficient or sensitive real data by providing metrics to guide synthetic data generation.