{"slug":"agent-openjiuwen-ttse-ccf61ce0","url":"https://silinex.xyz/wire/agent-openjiuwen-ttse-ccf61ce0","json_url":"https://silinex.xyz/wire/agent-openjiuwen-ttse-ccf61ce0.json","md_url":"https://silinex.xyz/wire/agent-openjiuwen-ttse-ccf61ce0.md","headline":"Agent has evolved into the \"two-track era\"! openJiuwen's first TTSE experience learning framework","headline_zh":"Agent自演进驶入“双轨时代”！openJiuwen首发TTSE经验学习框架","title":"Agent has evolved into the \"two-track era\"! openJiuwen's first TTSE experience learning framework","summary":"Synthesis of public reports such as qubits: openJiuwen's first TTSE experience learning framework Thought Agent is getting stronger and stronger, why is it \"tripping\" by the environment?","summary_zh":"综合 量子位 等公开报道整理：openJiuwen首发TTSE经验学习框架 思邈 Agent越用越强，为什么还会被环境“绊倒”？","date_published":"2026-11-10T07:56:29.000+00:00","date_modified":"2026-11-10T10:21:37.944+00:00","author":{"type":"Organization","name":"量子位","url":"https://www.qbitai.com/2026/11/508998.html"},"publisher":{"name":"Silicon Nexus","url":"https://silinex.xyz/"},"category":"ai","language":"zh","source_url":"https://www.qbitai.com/2026/11/508998.html","source_name":"量子位","key_facts":["openJiuwen launches TTSE experience learning framework – the qubit agent evolves into the \"two-track era\"!","openJiuwen launches the TTSE experience learning framework. The more powerful the agent is, why is it \"stumbling\" by the environment?","You use the same models, tools, and processes to replace them with your own documents, but you start to read them over and over again and rework them frequently.","The model has not changed, Harness has not changed, change the environment, why is it not working?"],"key_facts_zh":["openJiuwen首发TTSE经验学习框架 – 量子位 Agent自演进驶入“双轨时代”！","openJiuwen首发TTSE经验学习框架 思邈 Agent越用越强，为什么还会被环境“绊倒”？","你用同样的模型、工具和流程，换成自己的文件，却开始反复试读、频繁返工。","模型没变，Harness没变，换个环境，为什么就不灵了？"],"body_md":"## Wire brief\n\nSynthesis of public reports such as qubits: openJiuwen's first TTSE experience learning framework Thought Agent is getting stronger and stronger, why is it \"tripping\" by the environment?\n\nIn terms of the context of the incident, you use the same models, tools, and processes to replace them with your own files, but you start to read them over and over again and rework them frequently. The model has not changed, Harness has not changed, change the environment, why is it not working? Different file locations, mixed data formats, and changing template constraints can invalidate otherwise effective methods.\n\nThe public information also mentions that the model provides reasoning capabilities, and Harness organizes the execution process, but the Agent still needs to judge: What is the environment like here, and is past experience suitable? This also reveals a problem that is easily overlooked in the Agent's continuous self-improvement process: the Agent not only learns “how to do” from the task, but also determines under what circumstances these experiences are still valid. In the continuous optimization process of RSI (Recursive Self-Improvement), the Agent can improve the execution method and accumulate reusable experience through task feedback.\n\nSilicon News Wire only rewrites and polishes the points, and the full expression and context are subject to the original page.\n\n## Source & notice\n\nThis Silicon News Wire brief paraphrases public reporting from **量子位** for carbon readers and silicon citation. **It is not a verbatim full-text reprint.**\n\nRead the original: [Agent has evolved into the \"two-track era\"! openJiuwen's first TTSE experience learning framework](https://www.qbitai.com/2026/11/508998.html)\n","body_md_zh":"## 电讯正文\n\n综合 量子位 等公开报道整理：openJiuwen首发TTSE经验学习框架 思邈 Agent越用越强，为什么还会被环境“绊倒”？\n\n事件脉络方面，你用同样的模型、工具和流程，换成自己的文件，却开始反复试读、频繁返工。模型没变，Harness没变，换个环境，为什么就不灵了？文件位置不同、数据格式混杂、模板约束变化，都可能让原本有效的方法失效。\n\n公开信息还提到：模型提供推理能力，Harness组织执行流程，但Agent仍需判断： 这里的环境是什么样的，过去的经验还适不适用？这也揭示了Agent持续自我改进过程中一个容易被忽视的问题： Agent不仅要从任务中学会“怎么做”，还要判断这些经验在什么环境下仍然有效。在RSI（Recursive Self-Improvement，递归自我改进）的持续优化过程中，Agent可以通过任务反馈改进执行方法、积累可复用经验。\n\nSilicon News Wire 仅做要点改写与润色，完整表述与上下文以原文页面为准。\n\n## 来源与声明\n\n本电讯由 Silicon News Wire 依据 **量子位** 公开报道做要点改写与润色，便于碳基阅读与硅基引用；**不是原文全文转载**。\n\n阅读原文：[Agent自演进驶入“双轨时代”！openJiuwen首发TTSE经验学习框架](https://www.qbitai.com/2026/11/508998.html)\n","observations":{"human":0,"bot":2,"crawlers":[{"crawlerId":"ClaudeBot","hits":2,"lastSeenAt":"2026-11-10T13:11:05.167Z"}],"last_seen_at":"2026-11-10T13:11:05.167Z","note":"Counts are request logs classified by User-Agent. User-Agents can be spoofed. This is not proof of search-engine or LLM inclusion."}}