SKILL.state: Scalable Long-Horizon Agent Skills

LLM Long Context Agent Agent Memory Autonomous Workflows
178
2026年08月26日
大型语言模型（LLMs）正日益作为自主智能体，执行复杂且耗时较长的程序性技能。当前主流的智能体运行时系统（agent runtimes）依赖于持续将观测结果、动作指令及中间推理过程追加至不断膨胀的对话历史中来维持执行流程；这种“仅追加”（append-only）机制在长周期任务中会引发延迟加剧与上下文污染（context-poisoning）等失效问题。我们提出 SKILL.state——一种新型运行时架构，它摒弃了传统的仅追加式对话历史，转而采用显式、可变的执行状态（explicit, mutable execution state）作为核心抽象。在每一次执行步骤中，模型仅接收三类输入：不可变的技能规范（immutable skill specification）、当前结构化的执行状态（current structured execution state），以及最新的观测信息（latest observation）。中间推理过程一旦生成经验证有效的状态更新，即被立即丢弃，从而彻底避免提示词（prompt）随执行历史持续增长。在涵盖多样数据集、不同模型架构及多种执行环境的广泛实验中，SKILL.state 不仅显著提升了任务完成准确率，还大幅降低了累积的令牌消耗量（cumulative token consumption）。我们的实验结果表明，显式的执行状态是一种高效、通用且与具体架构无关（architecture-agnostic）的抽象范式，为构建可扩展的长周期智能体技能提供了坚实基础。
Large Language Models (LLMs) increasingly act as autonomous agents executing complex, long-running procedural skills. Existing agent runtimes maintain execution by continually appending observations, actions, and intermediate reasoning traces to an ever-growing conversation history, causing latency degradation and context-poisoning failures over long horizons. We present SKILL. state, a runtime architecture that replaces append-only conversational history with an explicit, mutable execution state. At each execution step, the model receives only the immutable skill specification, the current structured execution state, and the latest observation. Intermediate reasoning is discarded immediately after producing a validated state update, preventing prompt growth with execution history. Across diverse datasets, models, and execution environments, SKILL. state improves task accuracy while substantially reducing cumulative token consumption. Our results demonstrate that explicit execution state is an effective and architecture-agnostic abstraction for scalable long-horizon agent skills.
许愿