发表机构
China Telecom Cloud（中国电信云）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究针对ReAct范式LLM智能体的效率问题，提出TRIAGE三级路由框架，通过复用历史轨迹实现三级查询处理，在安全监控和跨域实验中显著减少token消耗，形成效率提升的正反馈循环。
AI 中文摘要
基于ReAct范式的大语言模型（LLM）智能体在工具使用和任务执行方面展现出卓越能力。然而ReAct存在根本性效率问题：每次查询都会从头触发完整推理循环，相似查询会重复相同步骤而不利用历史经验。我们提出TRIAGE，一种通过复用历史执行轨迹减少token消耗的三级路由框架，其核心创新是TaaS（Trajectory-as-a-Skill，将历史执行轨迹抽象为可复用技能，实现“经验即服务”）。TRIAGE将查询分为三级：（1）直接复用——相同查询，消耗0个token；（2）技能替换——相似查询，通过确定性参数替换消耗0个token；（3）完整ReAct——新颖查询，自动存储以供未来复用。在1007个安全监控查询的大规模实验中，TRIAGE实现62.3%的token节省，其中56.0%的查询属于Level 2、5.5%属于Level 1，均零成本执行；在ToolBench（15个领域、345个查询）的跨域验证中，实现76.3%的token减少，证实语义路由的通用性。在线学习实验显示出从冷启动到成熟的演化：前100个查询内Level 2命中率从0%升至57%，平均token成本从198降至74.7。我们还提出自动技能提取机制，将高频轨迹模式提炼为确定性技能，形成“使用越多、效率越高”的正反馈循环。
英文摘要
Large Language Model (LLM) agents based on the ReAct paradigm have demonstrated remarkable capabilities in tool use and task execution. However, ReAct suffers from a fundamental efficiency problem: every query triggers a complete reasoning loop from scratch, and similar queries repeat identical steps without leveraging historical experience. We propose TRIAGE,a three-level routing framework that reduces token consumption by reusing historical execution trajectories. Its core innovation is TaaS (Trajectory-as-a-Skill), which abstracts historical execution trajectories into reusable skills, realizing 'experience as a service'. TRIAGE classifies queries into three levels: (1) Direct Reuse-identical queries, 0 tokens; (2) Skill Substitution-similar queries, 0 tokens via deterministic parameter substitution; (3) Full ReAct-novel queries, automatically stored for future reuse. In large-scale experiments on 1,007 security monitoring queries, TRIAGE achieves 62.3% token savings, with 56.0% of queries at Level 2 and 5.5% at Level 1, both executing at zero cost. Cross-domain validation on ToolBench (15 domains, 345 queries) achieves 76.3% token reduction, confirming the generalizability of semantic routing. An online learning experiment demonstrates cold-start-to-mature evolution: the L2 hit rate rises from 0% to 57% within the first 100 queries, and the average token cost drops from 198 to 74.7. We also propose an automatic Skill extraction mechanism that distills high-frequency trajectory patterns into deterministic Skills, creating a positive feedback loop of 'the more you use it, the more efficient it becomes'.
