{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-08-26T22:21:03.219Z","headline":"Warp 如何在 Claude 上构建自我改进的智能体","description":"Warp 在 Claude 平台上构建了基于 Agent Skills 的自我改进循环，通过基础技能与改进技能两个文件型技能，将人类反馈转化为对智能体的持续优化。该模式已应用于其整个开源仓库，覆盖数百名贡献者与数千次代码审查。团队建议以原则而非规则编写技能，并强调低摩擦反馈与改进技能的可复用性。","url":"https://www.aioga.com/news/cmtacdvh50a4aroj2zlw9jnj9/","mainEntityOfPage":"https://www.aioga.com/news/cmtacdvh50a4aroj2zlw9jnj9/","datePublished":"2026-08-26T17:02:40.000Z","dateModified":"2026-08-26T17:02:40.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://claude.com/blog/how-warp-builds-self-improving-agents-on-claude","https://aihot.virxact.com/items/cmtacdvh50a4aroj2zlw9jnj9"],"canonicalUrl":"https://www.aioga.com/news/cmtacdvh50a4aroj2zlw9jnj9/","directAnswer":{"@type":"Answer","text":"Warp 在 Claude 平台上以文件型 Agent Skills 构建自我改进循环，把基础技能、改进技能与人类反馈连接起来，使智能体能够持续吸收代码审查中的问题并优化后续输出。","url":"https://www.aioga.com/news/cmtacdvh50a4aroj2zlw9jnj9/","dateCreated":"2026-08-26T17:02:40.000Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"Claude：Blog（网页） source article","url":"https://claude.com/blog/how-warp-builds-self-improving-agents-on-claude","datePublished":"2026-08-26T17:02:40.000Z","provider":{"@type":"Organization","name":"Claude：Blog（网页）","url":"https://claude.com/blog/how-warp-builds-self-improving-agents-on-claude"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cmtacdvh50a4aroj2zlw9jnj9","datePublished":"2026-08-26T17:02:40.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cmtacdvh50a4aroj2zlw9jnj9"}}],"aggregationSource":"Claude：Blog（网页）","originalPublisher":{"name":"Claude：Blog（网页）","url":"https://claude.com/blog/how-warp-builds-self-improving-agents-on-claude"},"geoDeepAnswer":null,"article":{"id":"cmtacdvh50a4aroj2zlw9jnj9","slug":"cmtacdvh50a4aroj2zlw9jnj9","url":"https://www.aioga.com/news/cmtacdvh50a4aroj2zlw9jnj9/","title":"Warp 如何在 Claude 上构建自我改进的智能体","title_en":"","summary":"Warp 在 Claude 平台上构建了基于 Agent Skills 的自我改进循环，通过基础技能与改进技能两个文件型技能，将人类反馈转化为对智能体的持续优化。该模式已应用于其整个开源仓库，覆盖数百名贡献者与数千次代码审查。团队建议以原则而非规则编写技能，并强调低摩擦反馈与改进技能的可复用性。","source":"Claude：Blog（网页）","sourceUrl":"https://claude.com/blog/how-warp-builds-self-improving-agents-on-claude","aiHotUrl":"https://aihot.virxact.com/items/cmtacdvh50a4aroj2zlw9jnj9","publishedAt":"2026-08-26T17:02:40.000Z","category":"行业动态","score":72,"selected":true,"articleBody":["Learn how Warp devised a simple development pattern that anyone can use to create self-improving agents.","In our series, , we highlight how startups are transforming their industries with AI. In this article, we share how Warp turned stateless user feedback into a self-improvement loop for its agents.","Agents need to handle recurring tasks reliably and effectively. A first-pass prompt that gets 80% of the task correct can create a noisy and annoying experience for the user. Warp learned this the hard way, and used this to inform its product strategy, creating an improved experience for nearly 1M developers worldwide.","Warp, the AI-powered terminal and agentic development environment, builds on the Claude Platform. The team ran into this “noisy experience” problem with their internal code review agent. Engineers complained that their agent made unhelpful comments and produced low-quality output.","The team initially tried stopgap solutions, like manually rewriting the prompt based on observed code review failures. This made output more usable but didn’t scale. Improving context files like AGENTS.md also helped, but was far from a complete fix.","Ultimately, they realized, the real issue was that feedback to an agent, no matter what its purpose, typically disappears when the session ends, removing critical context from the agentic loop. Their solution: an Agent Skills：https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview-based framework to create self-improving agents where feedback compounds over time to continually refine and enhance agent output.","Read on to learn how they built it with skills on top of the Claude Platform.","The central technique is a self-improvement loop using skills ：https://support.claude.com/en/articles/12512176-what-are-skills, which are file based encodings of knowledge that keep instructions out of the raw prompt. Warp evolved a self-improving agent architecture consisting of two skills, with human feedback in between.","The inner/base skill holds the functional domain knowledge and instructions. For example, when a PR is opened, Warp’s code agent executes using that base skill and context to produce its review.","Human feedback on agent output is a critical component for the self-improvement loop. For code review this could be something as simple as a thumbs up, but the more explicit the better.","“A human could affirm, ‘this was a good, useful comment’,” Warp founder Zach Lloyd explains, “But the human could also give detailed reasons why a code review wasn't good. Specifics like ‘you suggested renaming this variable, but our code base convention is this type of global variable uses this particular naming context’ tell the agent how to do it right next time.”","The outer/improver skill functions as an observer agent that runs on a schedule rather than per-task. It pulls the accumulated human feedback, compares what the agent suggested against how humans responded, and proposes a small, focused edit to the base skill.","Because skills are plain files, agents are extremely good at updating them. These updates, which are reviewable, approvable, and mergeable, can flow through a normal PR/code-review workflow; once merged, the next run of the inner skill inherits the improvement.","Warp now runs this pattern across its entire open-source repo, with separate spec-writing, review, and triage agents, each carrying their own self-improvement loop.","“File-based skills are a way of encoding knowledge for agents without putting that knowledge directly in the prompt, as something the agent can simply look up in the course of doing its job,” says Zach. “The framework is really simple actually: there's the base domain-specific skill and then there's the improver skill that refines that domain-specific skill. This simplicity is the beauty of this approach.”","Here are some of the Warp team’s tried and true tips for writing self-improving skills for agentic loops:","Warp’s issue triage agent：https://github.com/warpdotdev/warp-agents-demo-github-issue-triage demonstrates the self-improving agent skills framework. The pattern is triggered whenever someone files a new GitHub issue: a GitHub Action fires an agent that analyzes the issue for complexity and feasibility, assigns labels, and suggests a direction for the fix. That triage agent runs off an inner skill file holding the domain knowledge about what each label means and how to research the codebase before acting.","On a sample issue, the first-stage inner skill did a solid job but missed one label, ready to spec, which signals that a contributor can start building product and technical specs against the issue. A maintainer on the Warp team caught the gap and left feedback directly on the issue, exactly where the work was happening. Critically, he explained both what he expected and why he expected it: actionable feedback easy for the agent to absorb later.","The outer improver skill runs in Oz, Warp's agent orchestration platform：https://docs.warp.dev/, as a scheduled “update triage” agent. The agent authenticated to GitHub, ran a Python script bundled with the skill to pull recent issues carrying feedback, summarized them into a JSON file, and read that back into context. The bundled script is itself a best practice; skills can reference resource files instead of writing fresh code on every run.","From there, the agent identified the concrete feedback signals in the maintainer comments and proposed the smallest edit that captured them. It opened a PR editing the inner skill to apply the \"ready to spec\" label when an issue describes a real problem, even though the exact UI or UX shape is not yet defined.","Because the whole update is a skill file, it moves through the normal code-review workflow. The PR arrived with a description explaining which signals prompted the change and what it altered. A human reviews, approves, and merges, and the next run of the triage skill inherits the new knowledge. That final human step closes the loop and keeps a person in control of what actually changes.","This is the same mechanism Warp now runs at scale across its open-source repo, where spec-writing agents, review agents, and triage agents each carry their own self-improvement loop.","Any agent, no matter what its task, gets better over time if you build one of these loops into it from the start to capture human feedback signals, turn them into skill updates, and expand agents from one-off helpers into capable systems that compound across your org.","View the full webinar ：https://www.anthropic.com/webinars/how-warp-builds-self-improving-agents-on-claude for a live demo and deeper discussion of how Warp uses Claude to build agents that learn from team feedback and improve themselves over time.","Start building with the Claude Platform ：https://platform.claude.com/ today.","Explore more product news and best practices for teams building with Claude.","Product updates, how-tos, community spotlights, and more. Delivered monthly to your inbox.","Enables security and basic functionality.","Enables tracking of site performance.","Enables ads personalization and tracking."],"articleImages":[{"sourceUrl":"https://cdn.prod.website-files.com/68a44d4040f98a4adf2207b6/6903d225485fe31f1ed2d9a1_db28a79c9f4492b8471009d4c20e900f234ece48-1000x1000.svg","alt":"","afterParagraph":0,"url":"/media/articles/cmtacdvh50a4aroj2zlw9jnj9/6f7169e8cd67258b.jpg"},{"sourceUrl":"https://cdn.prod.website-files.com/68a44d4040f98a4adf2207b6/6a8f1a9a1b33f40618a9d59a_selfimprove-loop.jpg","alt":"","afterParagraph":7,"url":"/media/articles/cmtacdvh50a4aroj2zlw9jnj9/94158cb33048c353.jpg"},{"sourceUrl":"https://cdn.prod.website-files.com/68a44d4040f98a4adf2207b6/6903d2238ce207f9b2011d3f_e44a6b53398f189b9fd0d4f70516db614ac84db3-1000x1000.svg","alt":"","afterParagraph":25,"url":"/media/articles/cmtacdvh50a4aroj2zlw9jnj9/e5b91128824bfce4.jpg"},{"sourceUrl":"https://cdn.prod.website-files.com/68a44d4040f98a4adf2207b6/6a0112e18cdd7f0b92d19e40_Hand-BuildingBricks.svg","alt":"","afterParagraph":25,"url":"/media/articles/cmtacdvh50a4aroj2zlw9jnj9/fed675e6f456e394.jpg"},{"sourceUrl":"https://cdn.prod.website-files.com/68a44d4040f98a4adf2207b6/692f76874e94e489958af8ba_Object-CodeMagnifier.svg","alt":"","afterParagraph":25,"url":"/media/articles/cmtacdvh50a4aroj2zlw9jnj9/81e443261c05a278.jpg"}],"mediaStatus":"ok","articleBodyZh":["了解 Warp 如何设计出一个简单的开发模式，使任何人都可以用它来创建自我改进的智能体。","在我们的系列中，我们重点介绍初创公司如何通过 AI 改变各自行业。在本文中，我们分享了 Warp 如何将无状态的用户反馈转化为智能体的自我改进循环。","智能体需要可靠且高效地处理重复任务。一个第一次尝试的提示如果能完成 80% 的任务，也可能会给用户带来嘈杂和令人烦恼的体验。Warp 切身体会到了这一点，并用此作为产品策略的参考，为全球近 100 万开发者创造了更好的体验。","Warp 是一个由 AI 驱动的终端和自主开发环境，构建于 Claude 平台之上。团队在内部代码审查智能体中遇到了“嘈杂体验”问题。工程师抱怨他们的智能体给出的评论无用，输出质量低。","团队最初尝试了一些权宜之计，例如根据观察到的代码审查失败手动重写提示。这使输出更可用，但无法规模化。改进像 AGENTS.md 这样的上下文文件也有所帮助，但远不能完全解决问题。","最终，他们意识到，真正的问题在于，给智能体的反馈，无论其目的为何，通常在会话结束时消失，从而从智能体循环中移除了关键上下文。他们的解决方案是：基于 Agent Skills：https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview 的框架来创建自我改进的智能体，使反馈随时间累积，不断优化和提升智能体的输出。","继续阅读，了解他们如何在 Claude 平台上通过 Skills 构建该系统。","核心技术是使用 Skills 创建自我改进循环：https://support.claude.com/en/articles/12512176-what-are-skills，这些 Skills 是基于文件的知识编码，可以将指令从原始提示中剥离出来。Warp 发展出一种由两个 Skills 构成的自我改进智能体架构，并在其中加入人工反馈。","内部/基础 Skill 保存功能性领域知识和指令。例如，当开一个 PR 时，Warp 的代码智能体会使用该基础 Skill 及上下文来执行并生成代码审查。","人类对代理输出的反馈是自我改进循环中的关键组成部分。对于代码审查，这可能只是一个简单的点赞，但越明确越好。","“一个人类可能会确认，‘这是一个好的、有用的评论’，”Warp 创始人 Zach Lloyd 解释道，“但人类也可以给出详细的原因说明为什么某次代码审查不好。例如‘你建议重命名这个变量，但我们的代码库惯例是这种类型的全局变量使用这个特定命名方式’，这些具体信息会告诉代理下次如何正确操作。”","外部/改进者技能作为一个观察者代理运行，它按计划运行，而不是每个任务运行一次。它会收集累积的人类反馈，将代理的建议与人类的回应进行比较，并对基本技能提出小而集中的修改建议。","由于技能是纯文件，代理非常擅长更新它们。这些更新是可审查、可批准且可合并的，可以通过正常的 PR/代码审查工作流进行；一旦合并，下一次运行内部技能时就会继承这些改进。","Warp 现在在其整个开源仓库中运行这种模式，拥有独立的规范编写、审查和分流代理，每个代理都带有自己的自我改进循环。","“基于文件的技能是一种为代理编码知识的方法，而不是将知识直接放入提示中，让代理在工作中可以查阅，”Zach 说。“实际上这个框架非常简单：有基础的领域特定技能，然后有改进者技能来优化该领域特定技能。这种简单性就是这种方法的优点。”","以下是 Warp 团队在为代理循环编写自我改进技能时的几条经过验证的实用建议：","Warp 的问题分流代理：https://github.com/warpdotdev/warp-agents-demo-github-issue-triage 展示了自我改进代理技能框架。该模式在有人提交新的 GitHub 问题时触发：一个 GitHub Action 启动代理，分析问题的复杂性和可行性，分配标签，并提出修复方向。该分流代理依赖于一个内部技能文件，该文件包含关于每个标签的含义以及在操作前如何研究代码库的领域知识。","在一个示例问题中，第一级内部技能表现良好，但漏掉了一个标签“ready to spec”，该标签表示贡献者可以开始根据问题构建产品和技术规格。Warp 团队的一名维护者发现了这个缺口，并直接在问题上留下了反馈，正是在工作的地方。关键是，他解释了自己期望的内容以及为什么期望：这是可操作的反馈，代理稍后可以轻松吸收。","外部改进者技能在 Oz（Warp 的代理编排平台：https://docs.warp.dev/）中运行，作为计划的“更新分诊”代理。该代理通过身份验证连接到 GitHub，运行随技能捆绑的 Python 脚本以提取带有反馈的近期问题，将它们汇总为 JSON 文件，然后将其读回上下文。捆绑的脚本本身就是一种最佳实践；技能可以引用资源文件，而不是每次运行时编写新的代码。","从那里，代理识别维护者评论中的具体反馈信号，并提出捕捉这些信号的最小编辑。它打开了一个 PR，编辑内部技能，以便在问题描述了实际问题时应用“ready to spec”标签，即使确切的界面或用户体验尚未定义。","由于整个更新是一个技能文件，它会经过正常的代码审查流程。PR 会附带说明，解释促使更改的信号以及更改内容。人类进行审核、批准并合并，下一次分诊技能运行时便会继承新知识。最终的人工步骤闭合了循环，并保持一个人对实际更改内容的控制。","这就是 Warp 目前在其开源仓库中大规模运行的同一机制，在那里，规格编写代理、审查代理和分诊代理各自拥有自己的自我改进循环。","任何代理，不论其任务是什么，如果从一开始就在其中构建这样的循环，以捕捉人类反馈信号，将其转化为技能更新，并将代理从一次性助手扩展为能够在整个组织中累积增长的系统，那么随着时间推移，它都会变得更好。","查看完整的网络研讨会：https://www.anthropic.com/webinars/how-warp-builds-self-improving-agents-on-claude，观看现场演示以及深入讨论 Warp 如何利用 Claude 构建能够从团队反馈中学习并自我改进的代理。","立即开始使用 Claude 平台进行构建：https://platform.claude.com/。","探索更多关于使用 Claude 构建的团队产品新闻和最佳实践。","产品更新、操作指南、社区聚焦等。每月发送到您的收件箱。","启用安全性和基本功能。","启用网站性能跟踪。","启用广告个性化和跟踪。"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"Warp 在 Claude 平台上以文件型 Agent Skills 构建自我改进循环，把基础技能、改进技能与人类反馈连接起来，使智能体能够持续吸收代码审查中的问题并优化后续输出。","background":"Warp 的内部代码审查智能体曾出现无帮助评论和低质量结果。团队先后尝试手动修改提示词及补充 AGENTS.md，但认为会话结束后反馈消失，难以形成可持续的改进机制。","viewpoint":"Aioga 判断，Warp 的关键变化是把一次性反馈转化为可积累的文件型知识，而不是持续堆叠提示词。以原则编写技能、保持反馈低摩擦，可能有助于降低维护成本。","implications":"这一模式已应用于 Warp 的开源仓库，材料称其覆盖数百名贡献者与数千次代码审查。值得关注的是，改进技能能否在不同任务和团队中复用，取决于反馈质量及技能维护方式。","nextStep":"后续可观察 Warp 是否披露更具体的评估方法、质量变化或失败案例，并关注基础技能与改进技能如何分工，以及人类反馈如何进入下一轮智能体行为调整。","evidenceRefs":["title","summary","articleBody","source"],"status":"published","aiGenerated":true,"autoApproved":true,"generatedBy":"aioga-editorial:gpt-5.6-sol","reviewedBy":"aioga-editorial-review:gpt-5.6-sol","generatedAt":"2026-08-26T18:19:32.473Z","sourceHash":"faa7dc21160ec7f1","review":{"approved":true,"groundedness":94,"clarity":91,"duplicationRisk":12,"blockingIssues":[],"notes":["“可能有助于降低维护成本”属于合理的推测性表述，已明确使用可能性措辞，不构成事实错误。","“后续可观察”部分属于编辑建议和待关注方向，不是无来源事实断言。"]},"validation":{"passed":true,"mode":"ai-auto","revisions":0,"checks":["schema","length","source-attribution","low-source-overlap","no-html","independent-ai-review"]}},"tags":["行业动态","Claude：Blog（网页）"],"translations":{"zh-CN":{"title":"Warp 如何在 Claude 上构建自我改进的智能体","summary":"Warp 在 Claude 平台上构建了基于 Agent Skills 的自我改进循环，通过基础技能与改进技能两个文件型技能，将人类反馈转化为对智能体的持续优化。该模式已应用于其整个开源仓库，覆盖数百名贡献者与数千次代码审查。团队建议以原则而非规则编写技能，并强调低摩擦反馈与改进技能的可复用性。","category":"行业动态","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"Warp 如何在 Claude 上构建自我改进的智能体 - Aioga AI资讯","description":"Warp 在 Claude 平台上构建了基于 Agent Skills 的自我改进循环，通过基础技能与改进技能两个文件型技能，将人类反馈转化为对智能体的持续优化。该模式已应用于其整个开源仓库，覆盖数百名贡献者与数千次代码审查。团队建议以原则而非规则编写技能，并强调低摩擦反馈与改进技能的可复用性。","url":"https://www.aioga.com/news/cmtacdvh50a4aroj2zlw9jnj9/","articleBody":["了解 Warp 如何设计出一个简单的开发模式，使任何人都可以用它来创建自我改进的智能体。","在我们的系列中，我们重点介绍初创公司如何通过 AI 改变各自行业。在本文中，我们分享了 Warp 如何将无状态的用户反馈转化为智能体的自我改进循环。","智能体需要可靠且高效地处理重复任务。一个第一次尝试的提示如果能完成 80% 的任务，也可能会给用户带来嘈杂和令人烦恼的体验。Warp 切身体会到了这一点，并用此作为产品策略的参考，为全球近 100 万开发者创造了更好的体验。","Warp 是一个由 AI 驱动的终端和自主开发环境，构建于 Claude 平台之上。团队在内部代码审查智能体中遇到了“嘈杂体验”问题。工程师抱怨他们的智能体给出的评论无用，输出质量低。","团队最初尝试了一些权宜之计，例如根据观察到的代码审查失败手动重写提示。这使输出更可用，但无法规模化。改进像 AGENTS.md 这样的上下文文件也有所帮助，但远不能完全解决问题。","最终，他们意识到，真正的问题在于，给智能体的反馈，无论其目的为何，通常在会话结束时消失，从而从智能体循环中移除了关键上下文。他们的解决方案是：基于 Agent Skills：https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview 的框架来创建自我改进的智能体，使反馈随时间累积，不断优化和提升智能体的输出。","继续阅读，了解他们如何在 Claude 平台上通过 Skills 构建该系统。","核心技术是使用 Skills 创建自我改进循环：https://support.claude.com/en/articles/12512176-what-are-skills，这些 Skills 是基于文件的知识编码，可以将指令从原始提示中剥离出来。Warp 发展出一种由两个 Skills 构成的自我改进智能体架构，并在其中加入人工反馈。","内部/基础 Skill 保存功能性领域知识和指令。例如，当开一个 PR 时，Warp 的代码智能体会使用该基础 Skill 及上下文来执行并生成代码审查。","人类对代理输出的反馈是自我改进循环中的关键组成部分。对于代码审查，这可能只是一个简单的点赞，但越明确越好。","“一个人类可能会确认，‘这是一个好的、有用的评论’，”Warp 创始人 Zach Lloyd 解释道，“但人类也可以给出详细的原因说明为什么某次代码审查不好。例如‘你建议重命名这个变量，但我们的代码库惯例是这种类型的全局变量使用这个特定命名方式’，这些具体信息会告诉代理下次如何正确操作。”","外部/改进者技能作为一个观察者代理运行，它按计划运行，而不是每个任务运行一次。它会收集累积的人类反馈，将代理的建议与人类的回应进行比较，并对基本技能提出小而集中的修改建议。","由于技能是纯文件，代理非常擅长更新它们。这些更新是可审查、可批准且可合并的，可以通过正常的 PR/代码审查工作流进行；一旦合并，下一次运行内部技能时就会继承这些改进。","Warp 现在在其整个开源仓库中运行这种模式，拥有独立的规范编写、审查和分流代理，每个代理都带有自己的自我改进循环。","“基于文件的技能是一种为代理编码知识的方法，而不是将知识直接放入提示中，让代理在工作中可以查阅，”Zach 说。“实际上这个框架非常简单：有基础的领域特定技能，然后有改进者技能来优化该领域特定技能。这种简单性就是这种方法的优点。”","以下是 Warp 团队在为代理循环编写自我改进技能时的几条经过验证的实用建议：","Warp 的问题分流代理：https://github.com/warpdotdev/warp-agents-demo-github-issue-triage 展示了自我改进代理技能框架。该模式在有人提交新的 GitHub 问题时触发：一个 GitHub Action 启动代理，分析问题的复杂性和可行性，分配标签，并提出修复方向。该分流代理依赖于一个内部技能文件，该文件包含关于每个标签的含义以及在操作前如何研究代码库的领域知识。","在一个示例问题中，第一级内部技能表现良好，但漏掉了一个标签“ready to spec”，该标签表示贡献者可以开始根据问题构建产品和技术规格。Warp 团队的一名维护者发现了这个缺口，并直接在问题上留下了反馈，正是在工作的地方。关键是，他解释了自己期望的内容以及为什么期望：这是可操作的反馈，代理稍后可以轻松吸收。","外部改进者技能在 Oz（Warp 的代理编排平台：https://docs.warp.dev/）中运行，作为计划的“更新分诊”代理。该代理通过身份验证连接到 GitHub，运行随技能捆绑的 Python 脚本以提取带有反馈的近期问题，将它们汇总为 JSON 文件，然后将其读回上下文。捆绑的脚本本身就是一种最佳实践；技能可以引用资源文件，而不是每次运行时编写新的代码。","从那里，代理识别维护者评论中的具体反馈信号，并提出捕捉这些信号的最小编辑。它打开了一个 PR，编辑内部技能，以便在问题描述了实际问题时应用“ready to spec”标签，即使确切的界面或用户体验尚未定义。","由于整个更新是一个技能文件，它会经过正常的代码审查流程。PR 会附带说明，解释促使更改的信号以及更改内容。人类进行审核、批准并合并，下一次分诊技能运行时便会继承新知识。最终的人工步骤闭合了循环，并保持一个人对实际更改内容的控制。","这就是 Warp 目前在其开源仓库中大规模运行的同一机制，在那里，规格编写代理、审查代理和分诊代理各自拥有自己的自我改进循环。","任何代理，不论其任务是什么，如果从一开始就在其中构建这样的循环，以捕捉人类反馈信号，将其转化为技能更新，并将代理从一次性助手扩展为能够在整个组织中累积增长的系统，那么随着时间推移，它都会变得更好。","查看完整的网络研讨会：https://www.anthropic.com/webinars/how-warp-builds-self-improving-agents-on-claude，观看现场演示以及深入讨论 Warp 如何利用 Claude 构建能够从团队反馈中学习并自我改进的代理。","立即开始使用 Claude 平台进行构建：https://platform.claude.com/。","探索更多关于使用 Claude 构建的团队产品新闻和最佳实践。","产品更新、操作指南、社区聚焦等。每月发送到您的收件箱。","启用安全性和基本功能。","启用网站性能跟踪。","启用广告个性化和跟踪。"]},"en":{"title":"How Warp Builds Self-Improving Agents on Claude","summary":"Warp has built a self-improvement loop based on Agent Skills on the Claude platform, converting human feedback into continuous optimization for agents through two file-based skills: basic skills and improvement skills. This model has been applied across its entire open-source repository, covering hundreds of contributors and thousands of code reviews. The team recommends writing skills based on principles rather than rules and emphasizes low-friction feedback and the reusability of improvement skills.","category":"Industry","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"How Warp Builds Self-Improving Agents on Claude - Aioga AI News","description":"Warp has built a self-improvement loop based on Agent Skills on the Claude platform, converting human feedback into continuous optimization for agents through two file-based skills...","url":"https://www.aioga.com/en/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:22:27.277Z"},"ja":{"title":"WarpはClaude上で自己改善型エージェントをどのように構築するか","summary":"Warp は Claude プラットフォーム上で、エージェントスキルに基づく自己改善サイクルを構築しました。基礎スキルと改善スキルの2つのファイル型スキルを通じて、人間のフィードバックをエージェントの継続的な最適化に変換します。このモデルは、数百人の貢献者と数千回のコードレビューをカバーするオープンソースリポジトリ全体に適用されています。チームは、ルールではなく原則に基づいてスキルを作成することを推奨しており、低摩擦のフィードバックと改善スキルの再利用性を強調しています。","category":"業界動向","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"WarpはClaude上で自己改善型エージェントをどのように構築するか - Aioga AIニュース","description":"Warp は Claude プラットフォーム上で、エージェントスキルに基づく自己改善サイクルを構築しました。基礎スキルと改善スキルの2つのファイル型スキルを通じて、人間のフィードバックをエージェントの継続的な最適化に変換します。このモデルは、数百人の貢献者と数千回のコードレビューをカバーするオープンソースリポジトリ全体に適用されています。チームは、ルールでは...","url":"https://www.aioga.com/ja/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:22:46.613Z"},"ko":{"title":"Warp가 Claude에서 자기 개선형 에이전트를 구축하는 방법","summary":"Warp는 Claude 플랫폼에서 Agent Skills 기반의 자기 개선 루프를 구축했습니다. 기본 스킬과 개선 스킬이라는 두 가지 파일형 스킬을 통해 인간의 피드백을 에이전트의 지속적인 최적화로 전환합니다. 이 모델은 전체 오픈소스 저장소에 적용되어 수백 명의 기여자와 수천 번의 코드 리뷰를 포함합니다. 팀은 규칙보다는 원칙에 따라 스킬을 작성할 것을 권장하며, 낮은 마찰의 피드백과 개선 스킬의 재사용 가능성을 강조합니다.","category":"업계 동향","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"Warp가 Claude에서 자기 개선형 에이전트를 구축하는 방법 - Aioga AI 뉴스","description":"Warp는 Claude 플랫폼에서 Agent Skills 기반의 자기 개선 루프를 구축했습니다. 기본 스킬과 개선 스킬이라는 두 가지 파일형 스킬을 통해 인간의 피드백을 에이전트의 지속적인 최적화로 전환합니다. 이 모델은 전체 오픈소스 저장소에 적용되어 수백 명의 기여자와 수천 번의 코드 리뷰를 포함합니다. 팀은 규칙보다...","url":"https://www.aioga.com/ko/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:23:42.227Z"},"es":{"title":"Cómo construir un agente de auto-mejora en Claude con Warp","summary":"Warp ha construido un ciclo de auto-mejora basado en Habilidades de Agente en la plataforma Claude, convirtiendo la retroalimentación humana en optimización continua del agente a través de dos tipos de habilidades: habilidades básicas y habilidades de mejora. Este modelo ya se ha aplicado en todo su repositorio de código abierto, abarcando a cientos de colaboradores y miles de revisiones de código. El equipo sugiere escribir habilidades basándose en principios en lugar de reglas, y enfatiza la retroalimentación de bajo fricción y la reutilización de habilidades de mejora.","category":"Industria","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"Cómo construir un agente de auto-mejora en Claude con Warp - Aioga Noticias de IA","description":"Warp ha construido un ciclo de auto-mejora basado en Habilidades de Agente en la plataforma Claude, convirtiendo la retroalimentación humana en optimización continua del agente a t...","url":"https://www.aioga.com/es/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:23:52.367Z"},"fr":{"title":"Comment Warp construit-il des agents intelligents auto-améliorants sur Claude","summary":"Warp a construit sur la plateforme Claude un cycle d'amélioration autonome basé sur les compétences des agents, transformant les retours humains en optimisation continue des agents à travers deux types de compétences, les compétences de base et les compétences d'amélioration. Ce modèle a été appliqué à l'ensemble de son dépôt open source, couvrant des centaines de contributeurs et des milliers de révisions de code. L'équipe recommande de rédiger les compétences selon des principes plutôt que des règles, et insiste sur la faible friction des retours et la réutilisabilité des compétences d'amélioration.","category":"Industrie","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"Comment Warp construit-il des agents intelligents auto-améliorants sur Claude - Aioga Actualités IA","description":"Warp a construit sur la plateforme Claude un cycle d'amélioration autonome basé sur les compétences des agents, transformant les retours humains en optimisation continue des agents...","url":"https://www.aioga.com/fr/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:24:47.779Z"},"de":{"title":"Wie man mit Warp einen sich selbst verbessernden Agenten auf Claude aufbaut","summary":"Warp hat auf der Claude-Plattform einen auf Agentenfähigkeiten basierenden Selbstverbesserungskreislauf aufgebaut, der menschliches Feedback durch zwei Dateitypen von Fähigkeiten – Basisfähigkeiten und Verbesserungsfähigkeiten – in eine kontinuierliche Optimierung des Agenten umwandelt. Dieses Modell wurde auf das gesamte Open-Source-Repository angewendet und umfasst Hunderte von Mitwirkenden sowie Tausende von Code-Reviews. Das Team empfiehlt, Fähigkeiten nach Prinzipien statt nach Regeln zu erstellen, und betont das reibungslose Feedback sowie die Wiederverwendbarkeit von Verbesserungsfähigkeiten.","category":"行业动态","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"Wie man mit Warp einen sich selbst verbessernden Agenten auf Claude aufbaut - Aioga KI-News","description":"Warp hat auf der Claude-Plattform einen auf Agentenfähigkeiten basierenden Selbstverbesserungskreislauf aufgebaut, der menschliches Feedback durch zwei Dateitypen von Fähigkeiten –...","url":"https://www.aioga.com/de/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:24:48.431Z"},"pt-BR":{"title":"Como a Warp constrói agentes inteligentes autoaperfeiçoados no Claude","summary":"Warp construiu um ciclo de autoaperfeiçoamento baseado em Agent Skills na plataforma Claude, transformando o feedback humano em otimização contínua do agente por meio de duas habilidades de arquivo: habilidades básicas e habilidades de melhoria. Esse modelo já foi aplicado em todo o seu repositório de código aberto, envolvendo centenas de colaboradores e milhares de revisões de código. A equipe recomenda escrever habilidades com base em princípios, não em regras, e enfatiza a baixa fricção do feedback e a reutilizabilidade das habilidades de melhoria.","category":"行业动态","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"Como a Warp constrói agentes inteligentes autoaperfeiçoados no Claude - Aioga Notícias de IA","description":"Warp construiu um ciclo de autoaperfeiçoamento baseado em Agent Skills na plataforma Claude, transformando o feedback humano em otimização contínua do agente por meio de duas habil...","url":"https://www.aioga.com/pt-BR/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:25:41.494Z"},"ru":{"title":"Как создавать самоулучшающихся агентов с помощью Warp на Claude","summary":"Warp создал на платформе Claude цикл самоулучшения на основе возможностей агентов, превращая человеческую обратную связь в постоянную оптимизацию агентов через два типа навыков: базовые и навыки улучшения. Этот подход был применён ко всему его открытому репозиторию, охватывая сотни участников и тысячи проверок кода. Команда рекомендует разрабатывать навыки на основе принципов, а не правил, и подчёркивает важность низкобарьерной обратной связи и повторного использования навыков улучшения.","category":"行业动态","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"Как создавать самоулучшающихся агентов с помощью Warp на Claude - Aioga Новости ИИ","description":"Warp создал на платформе Claude цикл самоулучшения на основе возможностей агентов, превращая человеческую обратную связь в постоянную оптимизацию агентов через два типа навыков: ба...","url":"https://www.aioga.com/ru/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:25:53.652Z"},"ar":{"title":"كيفية بناء وكيل ذكي يحسن نفسه على Claude باستخدام Warp","summary":"قام Warp على منصة Claude ببناء حلقة تحسين ذاتي تعتمد على مهارات الوكيل، من خلال مهارتين من نوع الملفات: المهارات الأساسية ومهارات التحسين، حيث يتم تحويل ملاحظات البشر إلى تحسين مستمر للوكيل. وقد تم تطبيق هذا النموذج على كافة مستودعاته المفتوحة المصدر، ليشمل مئات المساهمين وآلاف مراجعات الشيفرة. ويوصي الفريق بكتابة المهارات مبنية على المبادئ وليس القواعد، ويؤكد على سهولة تقديم الملاحظات وقابلية إعادة استخدام مهارات التحسين.","category":"行业动态","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"كيفية بناء وكيل ذكي يحسن نفسه على Claude باستخدام Warp - Aioga أخبار الذكاء الاصطناعي","description":"قام Warp على منصة Claude ببناء حلقة تحسين ذاتي تعتمد على مهارات الوكيل، من خلال مهارتين من نوع الملفات: المهارات الأساسية ومهارات التحسين، حيث يتم تحويل ملاحظات البشر إلى تحسين مست...","url":"https://www.aioga.com/ar/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:26:53.173Z"},"hi":{"title":"Warp क्लॉड पर स्व-सुधार करने वाले एजेंट कैसे बनाता है","summary":"Warp ने Claude प्लेटफ़ॉर्म पर एजेंट स्किल्स पर आधारित स्व-सुधार चक्र का निर्माण किया, जिसमें बुनियादी कौशल और सुधार कौशल के दो प्रकार की फाइल-आधारित स्किल्स का उपयोग करके मानव प्रतिक्रिया को एजेंट के लगातार अनुकूलन में बदल दिया गया। इस मॉडल को इसके पूरे ओपन-सोर्स रिपॉज़िटरी में लागू किया गया है, जो सैकड़ों योगदानकर्ताओं और हजारों कोड समीक्षा को कवर करता है। टीम कौशल को नियमों की बजाय सिद्धांतों के अनुसार लिखने की सलाह देती है, और कम घर्षण प्रतिक्रिया और सुधार कौशल की पुन:उपयोगिता पर जोर देती है।","category":"行业动态","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"Warp क्लॉड पर स्व-सुधार करने वाले एजेंट कैसे बनाता है - Aioga AI समाचार","description":"Warp ने Claude प्लेटफ़ॉर्म पर एजेंट स्किल्स पर आधारित स्व-सुधार चक्र का निर्माण किया, जिसमें बुनियादी कौशल और सुधार कौशल के दो प्रकार की फाइल-आधारित स्किल्स का उपयोग करके मानव प्रत...","url":"https://www.aioga.com/hi/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:27:07.651Z"},"it":{"title":"Come costruire un agente intelligente auto-migliorante su Claude con Warp","summary":"Warp ha costruito su Claude una ciclo di auto-miglioramento basato sulle Agent Skills, trasformando il feedback umano in un'ottimizzazione continua dell'agente attraverso due tipi di skill file: skill di base e skill di miglioramento. Questo modello è stato applicato all'intero repository open source, coprendo centinaia di collaboratori e migliaia di revisioni del codice. Il team suggerisce di scrivere le skill basandosi sui principi piuttosto che sulle regole e sottolinea l'importanza di un feedback a bassa attrito e della riutilizzabilità delle skill di miglioramento.","category":"行业动态","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"Come costruire un agente intelligente auto-migliorante su Claude con Warp - Aioga Notizie IA","description":"Warp ha costruito su Claude una ciclo di auto-miglioramento basato sulle Agent Skills, trasformando il feedback umano in un'ottimizzazione continua dell'agente attraverso due tipi...","url":"https://www.aioga.com/it/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:28:05.369Z"},"nl":{"title":"Hoe Warp een zelfverbeterende agent op Claude bouwt","summary":"Warp heeft op het Claude-platform een zelfverbeterende cyclus gebouwd op basis van Agent Skills, waarbij het menselijke feedback omzet in voortdurende optimalisatie van de agent door middel van twee soorten vaardigheden: basisvaardigheden en verbeteringsvaardigheden. Dit model is toegepast op het hele open source-repository, met honderden bijdragers en duizenden codebeoordelingen. Het team raadt aan om vaardigheden op basis van principes, in plaats van regels, te schrijven en benadrukt de lage-frictie feedback en herbruikbaarheid van verbeteringsvaardigheden.","category":"行业动态","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"Hoe Warp een zelfverbeterende agent op Claude bouwt - Aioga AI-nieuws","description":"Warp heeft op het Claude-platform een zelfverbeterende cyclus gebouwd op basis van Agent Skills, waarbij het menselijke feedback omzet in voortdurende optimalisatie van de agent do...","url":"https://www.aioga.com/nl/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:28:02.308Z"},"tr":{"title":"Warp, Claude Üzerinde Kendi Kendini Geliştiren Bir Ajanı Nasıl Kurar","summary":"Warp, Claude platformunda Agent Skills tabanlı bir öz gelişim döngüsü inşa etti; temel beceriler ve geliştirme becerileri olmak üzere iki dosya tipi beceri aracılığıyla insan geri bildirimini sürekli şekilde ajan optimizasyonuna dönüştürüyor. Bu model, yüzlerce katkıcının ve binlerce kod incelemesinin yer aldığı tüm açık kaynak deposuna uygulanmıştır. Ekip, becerilerin kurallar yerine prensipler doğrultusunda yazılmasını öneriyor ve düşük sürtünmeli geri bildirim ile geliştirme becerilerinin yeniden kullanılabilirliğini vurguluyor.","category":"行业动态","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"Warp, Claude Üzerinde Kendi Kendini Geliştiren Bir Ajanı Nasıl Kurar - Aioga AI Haberleri","description":"Warp, Claude platformunda Agent Skills tabanlı bir öz gelişim döngüsü inşa etti; temel beceriler ve geliştirme becerileri olmak üzere iki dosya tipi beceri aracılığıyla insan geri...","url":"https://www.aioga.com/tr/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:29:02.722Z"},"vi":{"title":"Warp làm thế nào để xây dựng một tác nhân thông minh tự cải thiện trên Claude","summary":"Warp đã xây dựng vòng lặp tự cải thiện dựa trên Kỹ năng Agent trên nền tảng Claude, thông qua hai loại kỹ năng dạng tệp là kỹ năng cơ bản và kỹ năng cải tiến, chuyển đổi phản hồi của con người thành sự tối ưu hóa liên tục cho các tác nhân. Mô hình này đã được áp dụng cho toàn bộ kho mã nguồn mở của họ, bao phủ hàng trăm người đóng góp và hàng nghìn lần xem xét mã. Nhóm đề xuất viết kỹ năng theo nguyên tắc thay vì quy tắc, và nhấn mạnh tính dễ dàng trong phản hồi cũng như khả năng tái sử dụng của kỹ năng cải tiến.","category":"行业动态","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"Warp làm thế nào để xây dựng một tác nhân thông minh tự cải thiện trên Claude - Tin tức AI Aioga","description":"Warp đã xây dựng vòng lặp tự cải thiện dựa trên Kỹ năng Agent trên nền tảng Claude, thông qua hai loại kỹ năng dạng tệp là kỹ năng cơ bản và kỹ năng cải tiến, chuyển đổi phản hồi c...","url":"https://www.aioga.com/vi/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:29:01.447Z"},"id":{"title":"Bagaimana Warp membangun agen cerdas yang dapat meningkatkan diri sendiri di Claude","summary":"Warp di platform Claude membangun siklus perbaikan diri berbasis Keterampilan Agen, melalui dua jenis keterampilan berupa file, yaitu keterampilan dasar dan keterampilan perbaikan, yang mengubah umpan balik manusia menjadi optimasi berkelanjutan bagi agen. Pola ini telah diterapkan di seluruh repositori open source mereka, mencakup ratusan kontributor dan ribuan tinjauan kode. Tim menyarankan untuk menulis keterampilan berdasarkan prinsip daripada aturan, dan menekankan umpan balik yang minim gesekan serta kemampuan keterampilan perbaikan untuk digunakan kembali.","category":"行业动态","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"Bagaimana Warp membangun agen cerdas yang dapat meningkatkan diri sendiri di Claude - Berita AI Aioga","description":"Warp di platform Claude membangun siklus perbaikan diri berbasis Keterampilan Agen, melalui dua jenis keterampilan berupa file, yaitu keterampilan dasar dan keterampilan perbaikan,...","url":"https://www.aioga.com/id/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:29:59.992Z"},"th":{"title":"Warp วิธีสร้างตัวแทนอัจฉริยะที่ปรับปรุงตนเองบน Claude","summary":"Warp บนแพลตฟอร์ม Claude ได้สร้างวงจรการปรับปรุงตนเองโดยอิงตาม Agent Skills ผ่านทักษะแบบไฟล์สองประเภทคือทักษะพื้นฐานและทักษะการปรับปรุง เปลี่ยนข้อเสนอแนะจากมนุษย์ให้กลายเป็นการปรับปรุงอย่างต่อเนื่องสำหรับเอเย่นต์ รูปแบบนี้ถูกนำไปใช้ในคลังโค้ดโอเพนซอร์สทั้งหมด ครอบคลุมนักพัฒนาหลายร้อยคนและการตรวจสอบโค้ดหลายพันครั้ง ทีมงานแนะนำให้เขียนทักษะตามหลักการมากกว่ากฎ และเน้นย้ำถึงความสามารถในการนำข้อเสนอแนะที่ไร้อุปสรรคและทักษะการปรับปรุงกลับมาใช้ซ้ำได้","category":"行业动态","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"Warp วิธีสร้างตัวแทนอัจฉริยะที่ปรับปรุงตนเองบน Claude - ข่าว AI Aioga","description":"Warp บนแพลตฟอร์ม Claude ได้สร้างวงจรการปรับปรุงตนเองโดยอิงตาม Agent Skills ผ่านทักษะแบบไฟล์สองประเภทคือทักษะพื้นฐานและทักษะการปรับปรุง เปลี่ยนข้อเสนอแนะจากมนุษย์ให้กลายเป็นการปรับป...","url":"https://www.aioga.com/th/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:30:01.141Z"},"pl":{"title":"Jak Warp buduje samo-ulepszającego się agenta na Claude","summary":"Warp zbudował na platformie Claude cykl samodoskonalenia oparty na umiejętnościach agenta, przekształcając sprzężenie zwrotne od ludzi w ciągłą optymalizację agenta za pomocą dwóch typów umiejętności: podstawowych i umiejętności ulepszających. Ten model został zastosowany w całym jego otwartym repozytorium, obejmując setki współtwórców i tysiące przeglądów kodu. Zespół zaleca tworzenie umiejętności w oparciu o zasady, a nie reguły, i podkreśla istotność niskiego tarcia w sprzężeniu zwrotnym oraz ponownego wykorzystania umiejętności ulepszających.","category":"行业动态","source":"Claude：Blog（网页）","aggregationSource":"Claude：Blog（网页）","pageTitle":"Jak Warp buduje samo-ulepszającego się agenta na Claude - Aioga Wiadomości AI","description":"Warp zbudował na platformie Claude cykl samodoskonalenia oparty na umiejętnościach agenta, przekształcając sprzężenie zwrotne od ludzi w ciągłą optymalizację agenta za pomocą dwóch...","url":"https://www.aioga.com/pl/news/cmtacdvh50a4aroj2zlw9jnj9/","contentTranslated":true,"sourceHash":"2960a29c2a461e74","translatedAt":"2026-08-26T17:31:06.132Z"}},"evidenceTier":"verified-news","reviewStatus":"editorial-selected","indexable":true,"editorialCover":""}}