Recuris：双记忆架构提升长程智能体任务成功率
If you maintain a skill library for long-horizon agents, this one is worth your…
做长程 Agent 的同学建议收藏，这套双记忆架构给出了可落地的技能库维护思路，实测提升明显，值得在你的任务上复现验证。
If you maintain a skill library for long-horizon agents, this one is worth your time.
如果你为长时程智能体维护技能库，这篇值得你花时间阅读。
(bookmark it)
（记得收藏）
It discusses one of most common topics I get asked about these days.
它讨论了近期我最常被问及的话题之一。
It shares some good ideas on how to effectively leverage memory to improve the effectiveness of long-horizon agents.
它分享了一些关于如何有效利用记忆来提升长时程智能体效能的好点子。
Recuris splits agent memory in two. A Working Memory tracks task progress, and an Experiential Memory holds skills.
Recuris将智能体记忆分为两部分。工作记忆追踪任务进度，经验记忆存储技能。
Skill selection is grounded in the current task state instead of the full growing history, which is where long runs usually fall apart.
技能选择基于当前任务状态而非不断增长的历史记录，这正是长时运行通常失败之处。
Because skill use is anchored to an explicit state, a failed run points at a specific memory component. A fixed Meta-Agent turns the evidence into validation-gated updates to Skill Memory, which reshape execution and produce new evidence.
由于技能使用锚定于明确状态，失败运行能指向特定记忆组件。固定的元智能体将证据转化为经验证门控的技能记忆更新，这些更新重塑执行并产生新证据。
Across four long-horizon benchmarks and ten models, it improves task success in 35 of 37 completed model-benchmark pairs. On tau-bench it adds 17.8 points to GPT-5.6 Sol and 15.6 points to Claude Opus 5, taking Opus 5 to 87.9 percent.
在四个长时程基准和十个模型上，它在37个完成的模型-基准配对中提升了35个的任务成功率。在tau-bench上，它为GPT-5.6 Sol增加了17.8个百分点，为Claude Opus 5增加了15.6个百分点，使Opus 5达到87.9%。
The advantage widens as the horizon grows, reaching 32.2 points on the longest tasks. Common long-horizon failures drop by up to 80 percent.
随着时程增长，优势扩大，在最长任务上达到32.2个百分点。常见的长时程失败减少高达80%。
Paper: https://arxiv.org/abs/2608.24876
论文：https://arxiv.org/abs/2608.24876
Chat with Paper: https://academy.dair.ai/papers/recuris-evolves-working-and-experiential-memory-for-long-horizon-agents-2608.24876
与论文对话：https://academy.dair.ai/papers/recuris-evolves-working-and-experiential-memory-for-long-horizon-agents-2608.24876
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力