{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-10-23T01:00:39.594Z","headline":"Agent 不需要记忆插件，需要文档式记忆","description":"作者 kmeh 撰文认为市面上的 Agent 记忆插件架构错误，它们普遍用会话记录生成片段存入 RAG 数据库，每次提示词检索注入最相似的片段，但存在相似度检索无法判断正确性、片段丢失上下文、无法审计等问题。","url":"https://www.aioga.com/news/oactlhauxke4xq1kl9xt97qol/","mainEntityOfPage":"https://www.aioga.com/news/oactlhauxke4xq1kl9xt97qol/","datePublished":"2026-10-04T09:46:05.000Z","dateModified":"2026-10-04T09:46:05.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://liao.gg/blog/agents-dont-need-memory","https://aihot.news/items/oactlhauxke4xq1kl9xt97qol"],"canonicalUrl":"https://www.aioga.com/news/oactlhauxke4xq1kl9xt97qol/","directAnswer":{"@type":"Answer","text":"作者 kmeh 认为，现有 Agent 记忆插件通常把会话生成的片段存入向量数据库，再按提示词检索相似片段注入上下文；该做法可能缺少语境，也难以审计。","url":"https://www.aioga.com/news/oactlhauxke4xq1kl9xt97qol/","dateCreated":"2026-10-04T09:46:05.000Z","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":"liao.gg source article","url":"https://liao.gg/blog/agents-dont-need-memory","datePublished":"2026-10-04T09:46:05.000Z","provider":{"@type":"Organization","name":"liao.gg","url":"https://liao.gg/blog/agents-dont-need-memory"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.news/items/oactlhauxke4xq1kl9xt97qol","datePublished":"2026-10-04T09:46:05.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.news/items/oactlhauxke4xq1kl9xt97qol"}}],"aggregationSource":"Hacker News 热门","originalPublisher":{"name":"liao.gg","url":"https://liao.gg/blog/agents-dont-need-memory"},"geoDeepAnswer":null,"article":{"id":"oactlhauxke4xq1kl9xt97qol","slug":"oactlhauxke4xq1kl9xt97qol","url":"https://www.aioga.com/news/oactlhauxke4xq1kl9xt97qol/","title":"Agent 不需要记忆插件，需要文档式记忆","title_en":"","summary":"作者 kmeh 撰文认为市面上的 Agent 记忆插件架构错误，它们普遍用会话记录生成片段存入 RAG 数据库，每次提示词检索注入最相似的片段，但存在相似度检索无法判断正确性、片段丢失上下文、无法审计等问题。","source":"Hacker News 热门","sourceUrl":"https://liao.gg/blog/agents-dont-need-memory","aiHotUrl":"https://aihot.news/items/oactlhauxke4xq1kl9xt97qol","publishedAt":"2026-10-04T09:46:05.000Z","category":"行业动态","score":58,"selected":false,"articleBody":["A memory plugin analyzes your conversations. It generates 1,000 isolated snippets and inserts them into a vector database. With your every prompt, it attaches the five most similar snippets; if the agent is confused (which it is), it manually searches for more. That’s the product they call “memory.”","It’s a strange thing to use when you think about what problem you’re trying to solve. You want your agent to understand your project. To know where a feature is, why it was built, what you agreed on and what you care about. Instead, you get a lottery over RAG snippets, injected on every prompt, hoping the right ones float up.","Even when it does work, the agent still doesn’t understand your project. The entire memory plugin ecosystem is solving the wrong problem.","Because agents don’t need memory. They need documentation.","Every memory plugin on the market works the same way:","That’s the whole architecture. Some tools are extra fancy; they let the agent search through past transcripts word for word. Or they implement some kind of multi-tier memory system that classifies short or long-term memory. Or they add a bunch of background daemons to review, merge, or deduplicate memories. “Dreamers” that rewrite memories overnight. Continuous context compression. Rerankers. Et cetera.","Each plugin tries to add new token-burning “features” to fix the same flawed architecture underneath. And that’s why none of them reliably work.","All of these memory plugins suffer from the same broad set of problems.","These are only five of the many problems that memory plugins face, attempt, and fail to fix, because they all make the same assumption:","Agents forget: that’s the problem. So the fix is to remember. To remember better, we should capture more, index better, retrieve smarter.","Their entire thesis revolves around capturing and recalling the past. But that’s not how anyone else handles knowledge. No one rewatches a team meeting from 3 years ago to remember constraints around a feature. People write things down and use those records instead.","Similarly, the solution is NOT to give an agent a search tool that it uses to search over the past 10 million tokens worth of conversations, reconstructing fragments of what happens.","The solution, is document-based memory.","Today, people use AI to pump out products, features, and slop at the speed of light, all without reading or understanding a single line of code. Under these circumstances, it’s easy to see how documentation ends up becoming an afterthought when it should be more important than ever.","People already knew agents needed context. That’s why they invented AGENTS.md files: so agents don’t jump into a codebase blind. It works. But oftentimes, that single file is the ONLY documentation that the project has.","A single file is not enough. The agent needs an entire brain, a structured workspace where it can record instructions, specs, decisions, research, indexes, anything without being asked. Instructions for how the code review process works, specs detailing what was discussed with the user, reusable research into a foreign library or API.","When the agent works, the agent can read files from the brain to gain relevant and complete context. After the agent works, it updates what’s outdated, adding new documents where necessary, while the full picture is still in context. In this way, the agentic loop changes from prompt → build → forget to prompt → consult → build → update . Memory turns from a RAG database you bolt onto the agent into a workspace you can read, update, and even share.","I recognized this problem over a year ago when I first started programming with AI. I wanted a way for an agent to remember work across sessions, so I started by creating an internal/ folder where I asked the agent to write everything down: specs, plans, indexes. I tasked the agent with always reading the appropriate documents and index before doing work and updating afterwards.","This set of rudimentary instructions slowly transformed into a formal system, and then into a plugin called Operator Memory that I’ve been using regularly within all of my projects.","Operator Memory provides document-based memory using the model described above. Operator provides your agent with a Markdown brain where it can persist important knowledge: instructions, specs, research, indexes. Before working, the agent always consults the brain for relevant documents. After working, the agent updates the brain; revisiting stale documents, adding documents where missing.","No vector databases. No embeddings. No summarizers, curators, updaters, dreamers, or any other token-burning background daemon. No black-box retrieval.","With Operator, everything is a plain Markdown document that you can read, update, commit, and share with your team. I’ve been using this system for over a year. If you want to try it out, it’s free and open source: https://github.com/aerovato/operator-memory：https://github.com/aerovato/operator-memory"],"articleImages":[{"sourceUrl":"https://liao.gg/images/blog/agents-dont-need-memory/tangled-1600.jpg","alt":"A tangle of colorful wires","afterParagraph":0,"url":"/media/articles/oactlhauxke4xq1kl9xt97qol/5ccb0161a4b43f46.jpg"},{"sourceUrl":"https://liao.gg/images/blog/agents-dont-need-memory/change-the-loop.png","alt":"Diagram comparing the prompt, build, forget loop with Operator Memory's prompt, consult, build, update loop","afterParagraph":18,"url":"/media/articles/oactlhauxke4xq1kl9xt97qol/209715c61cf26a01.png"}],"mediaStatus":"ok","articleBodyZh":["一个记忆插件会分析你的对话。它会生成1,000个独立片段并将它们插入向量数据库。每当你输入提示时，它会附上五个最相似的片段；如果智能体困惑（通常确实如此），它会手动搜索更多。那就是他们所谓的“记忆”产品。","当你考虑你想解决的问题时，这是一件奇怪的事情。你希望你的智能体理解你的项目。知道某个功能在哪里，为什么被构建，以及你们达成了什么共识，你在意什么。相反，你得到的是一个关于RAG片段的“彩票”，每次输入提示时注入，希望正确的片段浮现出来。","即便它确实有效，智能体仍然无法理解你的项目。整个记忆插件生态系统解决的是错误的问题。","因为智能体不需要记忆。它们需要的是文档。","市面上所有的记忆插件都是这样工作的：","这就是整个架构。一些工具很高级；它们让智能体逐字搜索过去的记录。或者它们实现某种多层记忆系统，将短期或长期记忆分类。或者它们添加了一堆后台守护程序来审查、合并或去重记忆。“梦想者”可以整夜重写记忆。持续的上下文压缩。重新排序器。等等。","每个插件都尝试添加新的消耗token的“功能”来修复底层相同的架构缺陷。这就是为什么它们都无法可靠工作。","所有这些记忆插件都有一组相同的广泛问题。","以下只是记忆插件面临、尝试并未能解决的众多问题中的五个，因为它们都有一个相同的假设：","智能体会忘记：这是问题所在。所以解决方法是记住。为了更好地记住，我们应该捕捉更多，更好地索引，更智能地检索。","它们的整个理论核心都是围绕捕捉和回忆过去。但是没有其他人是这样处理知识的。没人会重看三年前的团队会议来记住某个功能的限制。人们会写下来，然后使用这些记录。","同样，解决方案也不是给智能体一个搜索工具，让它用来搜索过去一千万token的对话片段，并重建发生的片段。","解决方案是基于文档的记忆。","如今，人们利用 AI 以光速推出产品、功能和杂乱内容，而不需要阅读或理解一行代码。在这种情况下，很容易理解为什么文档最终会被忽视，而它本应比以往任何时候都重要。","人们早就知道智能体需要上下文。这就是他们发明 AGENTS.md 文件的原因：让智能体不会盲目进入代码库。这确实有效。但很多时候，这个文件是项目唯一的文档。","一个文件是不够的。智能体需要一个完整的大脑，一个结构化的工作空间，可以无需请求就记录指令、规格、决策、研究、索引等。包括代码审查流程的操作指令、与用户讨论细节的规格、对外部库或 API 的可重复使用研究。","当智能体工作时，它可以从大脑中读取文件以获得相关且完整的上下文。工作完成后，它会更新过时内容，必要时添加新文档，同时保持完整的上下文。以这种方式，智能体循环从 prompt → build → forget 变为 prompt → consult → build → update。记忆从你附加到智能体的 RAG 数据库，变成一个你可以阅读、更新甚至分享的工作空间。","我在一年多前首次使用 AI 编程时就意识到了这个问题。我希望找到一种方法，让智能体可以跨会话记住工作内容，于是我首先创建了一个 internal/ 文件夹，让智能体记录所有内容：规格、计划、索引。我要求智能体在工作前总是阅读相关文档和索引，工作后进行更新。","这套初步指令慢慢演变成了正式系统，然后变成了一个名为 Operator Memory 的插件，我一直在我的所有项目中定期使用它。","Operator Memory 提供基于文档的记忆，使用上述模型。Operator 为你的智能体提供了一个 Markdown 大脑，它可以持久保存重要知识：指令、规格、研究、索引。工作前，智能体总是查阅大脑中的相关文档。工作后，智能体会更新大脑：重新审视过时文档，补充缺失文档。","没有向量数据库。没有嵌入。没有总结器、策展人、更新器、梦想家或任何其他消耗代币的后台守护程序。没有黑箱检索。","使用 Operator，一切都是可以阅读、更新、提交并与团队共享的普通 Markdown 文档。我已经使用这个系统一年多了。如果你想尝试，它是免费的且开源的： https://github.com/aerovato/operator-memory"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"作者 kmeh 认为，现有 Agent 记忆插件通常把会话生成的片段存入向量数据库，再按提示词检索相似片段注入上下文；该做法可能缺少语境，也难以审计。","background":"文章将相关插件概括为从对话生成孤立片段，并在每次提示时附加若干相似内容。作者认为，用户真正希望 Agent 理解项目结构、功能缘由、已达成共识及关注事项。","viewpoint":"Aioga 判断：文章的核心主张是，Agent 的长期理解不应主要依赖零散记忆片段，而应依赖能够呈现项目背景、决策依据和约定内容的文档。该判断建立在作者对检索局限的批评上。","implications":"可能影响：如果采用文档式记录，项目背景和决策内容可能更便于人工查看与追溯；但文章不足以证明文档方案在所有 Agent 场景中都优于记忆插件，具体效果仍需要实际验证。","nextStep":"后续观察：应关注相关工具是否提供清晰的文档组织、上下文维护和审计方式，也需要观察这些方式能否减少片段检索造成的语境缺失，并支持 Agent 理解项目约定。","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":"2026-10-04T13:47:00.860Z","sourceHash":"6328f10373c85c09","review":{"approved":true,"groundedness":94,"clarity":92,"duplicationRisk":18,"blockingIssues":[],"notes":["“Aioga 判断”属于编辑者归纳，可考虑统一为“文章的核心主张是”，以减少主体切换。","“文档式记录可能更便于人工查看与追溯”是合理推论，但并非来源明确验证的结论；当前已使用“可能”限定，建议保留该限定。"]},"validation":{"passed":true,"mode":"ai-auto","revisions":0,"checks":["schema","length","source-attribution","editorial-labels","inference-boundary","low-source-overlap","no-html","independent-ai-review"]}},"tags":["行业动态","Hacker News 热门"],"translations":{"zh-CN":{"title":"Agent 不需要记忆插件，需要文档式记忆","summary":"作者 kmeh 撰文认为市面上的 Agent 记忆插件架构错误，它们普遍用会话记录生成片段存入 RAG 数据库，每次提示词检索注入最相似的片段，但存在相似度检索无法判断正确性、片段丢失上下文、无法审计等问题。","category":"行业动态","source":"liao.gg","aggregationSource":"Hacker News 热门","pageTitle":"Agent 不需要记忆插件，需要文档式记忆 - Aioga AI资讯","description":"作者 kmeh 撰文认为市面上的 Agent 记忆插件架构错误，它们普遍用会话记录生成片段存入 RAG 数据库，每次提示词检索注入最相似的片段，但存在相似度检索无法判断正确性、片段丢失上下文、无法审计等问题。","url":"https://www.aioga.com/news/oactlhauxke4xq1kl9xt97qol/","articleBody":["一个记忆插件会分析你的对话。它会生成1,000个独立片段并将它们插入向量数据库。每当你输入提示时，它会附上五个最相似的片段；如果智能体困惑（通常确实如此），它会手动搜索更多。那就是他们所谓的“记忆”产品。","当你考虑你想解决的问题时，这是一件奇怪的事情。你希望你的智能体理解你的项目。知道某个功能在哪里，为什么被构建，以及你们达成了什么共识，你在意什么。相反，你得到的是一个关于RAG片段的“彩票”，每次输入提示时注入，希望正确的片段浮现出来。","即便它确实有效，智能体仍然无法理解你的项目。整个记忆插件生态系统解决的是错误的问题。","因为智能体不需要记忆。它们需要的是文档。","市面上所有的记忆插件都是这样工作的：","这就是整个架构。一些工具很高级；它们让智能体逐字搜索过去的记录。或者它们实现某种多层记忆系统，将短期或长期记忆分类。或者它们添加了一堆后台守护程序来审查、合并或去重记忆。“梦想者”可以整夜重写记忆。持续的上下文压缩。重新排序器。等等。","每个插件都尝试添加新的消耗token的“功能”来修复底层相同的架构缺陷。这就是为什么它们都无法可靠工作。","所有这些记忆插件都有一组相同的广泛问题。","以下只是记忆插件面临、尝试并未能解决的众多问题中的五个，因为它们都有一个相同的假设：","智能体会忘记：这是问题所在。所以解决方法是记住。为了更好地记住，我们应该捕捉更多，更好地索引，更智能地检索。","它们的整个理论核心都是围绕捕捉和回忆过去。但是没有其他人是这样处理知识的。没人会重看三年前的团队会议来记住某个功能的限制。人们会写下来，然后使用这些记录。","同样，解决方案也不是给智能体一个搜索工具，让它用来搜索过去一千万token的对话片段，并重建发生的片段。","解决方案是基于文档的记忆。","如今，人们利用 AI 以光速推出产品、功能和杂乱内容，而不需要阅读或理解一行代码。在这种情况下，很容易理解为什么文档最终会被忽视，而它本应比以往任何时候都重要。","人们早就知道智能体需要上下文。这就是他们发明 AGENTS.md 文件的原因：让智能体不会盲目进入代码库。这确实有效。但很多时候，这个文件是项目唯一的文档。","一个文件是不够的。智能体需要一个完整的大脑，一个结构化的工作空间，可以无需请求就记录指令、规格、决策、研究、索引等。包括代码审查流程的操作指令、与用户讨论细节的规格、对外部库或 API 的可重复使用研究。","当智能体工作时，它可以从大脑中读取文件以获得相关且完整的上下文。工作完成后，它会更新过时内容，必要时添加新文档，同时保持完整的上下文。以这种方式，智能体循环从 prompt → build → forget 变为 prompt → consult → build → update。记忆从你附加到智能体的 RAG 数据库，变成一个你可以阅读、更新甚至分享的工作空间。","我在一年多前首次使用 AI 编程时就意识到了这个问题。我希望找到一种方法，让智能体可以跨会话记住工作内容，于是我首先创建了一个 internal/ 文件夹，让智能体记录所有内容：规格、计划、索引。我要求智能体在工作前总是阅读相关文档和索引，工作后进行更新。","这套初步指令慢慢演变成了正式系统，然后变成了一个名为 Operator Memory 的插件，我一直在我的所有项目中定期使用它。","Operator Memory 提供基于文档的记忆，使用上述模型。Operator 为你的智能体提供了一个 Markdown 大脑，它可以持久保存重要知识：指令、规格、研究、索引。工作前，智能体总是查阅大脑中的相关文档。工作后，智能体会更新大脑：重新审视过时文档，补充缺失文档。","没有向量数据库。没有嵌入。没有总结器、策展人、更新器、梦想家或任何其他消耗代币的后台守护程序。没有黑箱检索。","使用 Operator，一切都是可以阅读、更新、提交并与团队共享的普通 Markdown 文档。我已经使用这个系统一年多了。如果你想尝试，它是免费的且开源的： https://github.com/aerovato/operator-memory"]},"en":{"title":"Agents do not require memory plugins; they require document-based memory","summary":"Author kmeh wrote that the Agent memory plugin on the market has flawed architectures. They commonly generate fragments from session records and store them in the RAG database, injecting the most similar segments each time prompt searches, but face issues such as similarity search failing to judge correctness, fragments missing context, and unauditability.","category":"Industry","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"Agents do not require memory plugins; they require document-based memory - Aioga AI News","description":"Author kmeh wrote that the Agent memory plugin on the market has flawed architectures. They commonly generate fragments from session records and store them in the RAG database, inj...","url":"https://www.aioga.com/en/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:02:37.211Z"},"ja":{"title":"エージェントはメモリプラグインを必要としません。ドキュメントベースのメモリが必要です","summary":"著者のkmehは、市場に出ているAgentメモリプラグインのアーキテクチャに欠陥があると述べています。これらはセッションレコードから断片を生成し、RAGデータベースに保存し、プロンプト検索のたびに最も似たセグメントを注入しますが、類似性検索が正確性を判断できず、断片が文脈を欠いていること、監査不能といった問題に直面しています。","category":"業界動向","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"エージェントはメモリプラグインを必要としません。ドキュメントベースのメモリが必要です - Aioga AIニュース","description":"著者のkmehは、市場に出ているAgentメモリプラグインのアーキテクチャに欠陥があると述べています。これらはセッションレコードから断片を生成し、RAGデータベースに保存し、プロンプト検索のたびに最も似たセグメントを注入しますが、類似性検索が正確性を判断できず、断片が文脈を欠いていること、監査不能といった問題に直面しています。","url":"https://www.aioga.com/ja/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:02:36.717Z"},"ko":{"title":"에이전트는 메모리 플러그인이 필요하지 않습니다; 문서 기반 메모리가 필요합니다","summary":"저자 kmeh는 시장에 나와 있는 에이전트 메모리 플러그인의 아키텍처에 결함이 있다고 썼습니다. 이들은 일반적으로 세션 레코드에서 조각을 생성하여 RAG 데이터베이스에 저장하며, 프롬프트 검색 시 가장 유사한 구간을 주입하지만, 유사도 검색이 정확성을 판단하지 못하거나, 조각이 맥락이 누락되며, 감사가 불가능한 문제도 있습니다.","category":"업계 동향","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"에이전트는 메모리 플러그인이 필요하지 않습니다; 문서 기반 메모리가 필요합니다 - Aioga AI 뉴스","description":"저자 kmeh는 시장에 나와 있는 에이전트 메모리 플러그인의 아키텍처에 결함이 있다고 썼습니다. 이들은 일반적으로 세션 레코드에서 조각을 생성하여 RAG 데이터베이스에 저장하며, 프롬프트 검색 시 가장 유사한 구간을 주입하지만, 유사도 검색이 정확성을 판단하지 못하거나, 조각이 맥락이 누락되며, 감사가 불가능한 문제도 있...","url":"https://www.aioga.com/ko/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:02:37.554Z"},"es":{"title":"Los agentes no requieren plugins de memoria; requieren memoria basada en documentos","summary":"El autor kmeh escribió que el plugin de memoria Agent del mercado tiene arquitecturas defectuosas. Suelen generar fragmentos a partir de registros de sesión y almacenarlos en la base de datos RAG, inyectando los segmentos más similares cada vez que se realiza una búsqueda por prompt, pero se enfrentan a problemas como la búsqueda por similitud que no juzga la corrección, fragmentos que faltan contexto y no auditabilidad.","category":"Industria","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"Los agentes no requieren plugins de memoria; requieren memoria basada en documentos - Aioga Noticias de IA","description":"El autor kmeh escribió que el plugin de memoria Agent del mercado tiene arquitecturas defectuosas. Suelen generar fragmentos a partir de registros de sesión y almacenarlos en la ba...","url":"https://www.aioga.com/es/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:03:01.642Z"},"fr":{"title":"Les agents ne nécessitent pas de plugins mémoire ; ils nécessitent une mémoire basée sur des documents","summary":"L’auteur kmeh a écrit que le plugin mémoire Agent sur le marché possède des architectures défectueuses. Ils génèrent couramment des fragments à partir des enregistrements de session et les stockent dans la base de données RAG, injectant les segments les plus similaires à chaque recherche d’invites, mais rencontrent des problèmes tels que la recherche par similarité qui ne juge pas la véracité, des fragments manquant de contexte, et l’inauditabilité.","category":"Industrie","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"Les agents ne nécessitent pas de plugins mémoire ; ils nécessitent une mémoire basée sur des documents - Aioga Actualités IA","description":"L’auteur kmeh a écrit que le plugin mémoire Agent sur le marché possède des architectures défectueuses. Ils génèrent couramment des fragments à partir des enregistrements de sessio...","url":"https://www.aioga.com/fr/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:03:01.706Z"},"de":{"title":"Agenten benötigen keine Speicher-Plugins; sie benötigen dokumentbasierten Speicher","summary":"Der Autor kmeh schrieb, dass das auf dem Markt erhältliche Agent-Speicher-Plugin fehlerhafte Architekturen aufweist. Sie erzeugen häufig Fragmente aus Sitzungsdatensätzen und speichern sie in der RAG-Datenbank, wobei sie bei jeder Suchanweisung die ähnlichsten Segmente einfügen, aber sie haben Probleme wie das Versäumnis der Ähnlichkeitssuche, Fragmente ohne Kontext und Unprüfbarkeit zu beurteilen.","category":"行业动态","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"Agenten benötigen keine Speicher-Plugins; sie benötigen dokumentbasierten Speicher - Aioga KI-News","description":"Der Autor kmeh schrieb, dass das auf dem Markt erhältliche Agent-Speicher-Plugin fehlerhafte Architekturen aufweist. Sie erzeugen häufig Fragmente aus Sitzungsdatensätzen und speic...","url":"https://www.aioga.com/de/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:03:01.753Z"},"pt-BR":{"title":"Agentes não requerem plugins de memória; eles requerem memória baseada em documentos","summary":"O autor kmeh escreveu que o plugin de memória Agent no mercado possui arquiteturas falhas. Eles comumente geram fragmentos a partir de registros de sessão e os armazenam no banco de dados RAG, injetando os segmentos mais semelhantes a cada busca por prompt, mas enfrentam problemas como busca por similaridade que não avalia a correção, fragmentos sem contexto e inaudibilidade.","category":"行业动态","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"Agentes não requerem plugins de memória; eles requerem memória baseada em documentos - Aioga Notícias de IA","description":"O autor kmeh escreveu que o plugin de memória Agent no mercado possui arquiteturas falhas. Eles comumente geram fragmentos a partir de registros de sessão e os armazenam no banco d...","url":"https://www.aioga.com/pt-BR/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:03:25.319Z"},"ru":{"title":"Агентам не нужны плагины памяти; им нужна документная память","summary":"Автор KMEH писал, что плагин памяти Agent на рынке имеет несовершенные архитектуры. Они обычно генерируют фрагменты из записей сессий и хранят их в базе данных RAG, вставляя наиболее похожие сегменты при каждом запросе запросов, но сталкиваются с такими проблемами, как неспособность оценивать корректность по сходству, отсутствие контекста фрагментов и непроверимость.","category":"行业动态","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"Агентам не нужны плагины памяти; им нужна документная память - Aioga Новости ИИ","description":"Автор KMEH писал, что плагин памяти Agent на рынке имеет несовершенные архитектуры. Они обычно генерируют фрагменты из записей сессий и хранят их в базе данных RAG, вставляя наибол...","url":"https://www.aioga.com/ru/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:03:25.293Z"},"ar":{"title":"الوكلاء لا يحتاجون إلى إضافات الذاكرة؛ بل يتطلب ذاكرة قائمة على المستندات","summary":"كتب المؤلف kmeh أن إضافة ذاكرة Agent في السوق تعاني من بنى معيبة. عادة ما تولد أجزاء من سجلات الجلسات وتخزنها في قاعدة بيانات RAG، مع حقن الأجزاء الأكثر تشابها في كل مرة يتم فيها البحث في الطلب، لكنها تواجه مشاكل مثل فشل البحث في التشابه في تحديد الصحة، وفقدان السياق في الشظايا، وعدم قابلية التدقيق.","category":"行业动态","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"الوكلاء لا يحتاجون إلى إضافات الذاكرة؛ بل يتطلب ذاكرة قائمة على المستندات - Aioga أخبار الذكاء الاصطناعي","description":"كتب المؤلف kmeh أن إضافة ذاكرة Agent في السوق تعاني من بنى معيبة. عادة ما تولد أجزاء من سجلات الجلسات وتخزنها في قاعدة بيانات RAG، مع حقن الأجزاء الأكثر تشابها في كل مرة يتم فيها ا...","url":"https://www.aioga.com/ar/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:03:25.122Z"},"hi":{"title":"एजेंटों को मेमोरी प्लगइन्स की आवश्यकता नहीं होती है; उन्हें दस्तावेज़-आधारित मेमोरी की आवश्यकता होती है","summary":"लेखक kmeh ने लिखा है कि बाजार पर एजेंट मेमोरी प्लगइन में त्रुटिपूर्ण आर्किटेक्चर हैं। वे आमतौर पर सत्र रिकॉर्ड से टुकड़े उत्पन्न करते हैं और उन्हें आरएजी डेटाबेस में संग्रहीत करते हैं, हर बार शीघ्र खोजों में सबसे समान खंडों को इंजेक्ट करते हैं, लेकिन समानता खोज जैसे मुद्दों का सामना करना पड़ता है जो शुद्धता का न्याय करने में विफल रहता है, टुकड़े गायब संदर्भ और अलेखापरीक्षितता।","category":"行业动态","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"एजेंटों को मेमोरी प्लगइन्स की आवश्यकता नहीं होती है; उन्हें दस्तावेज़-आधारित मेमोरी की आवश्यकता होती है - Aioga AI समाचार","description":"लेखक kmeh ने लिखा है कि बाजार पर एजेंट मेमोरी प्लगइन में त्रुटिपूर्ण आर्किटेक्चर हैं। वे आमतौर पर सत्र रिकॉर्ड से टुकड़े उत्पन्न करते हैं और उन्हें आरएजी डेटाबेस में संग्रहीत करते...","url":"https://www.aioga.com/hi/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:03:48.877Z"},"it":{"title":"Gli Agent non richiedono plugin di memoria; richiedono memoria basata su documenti","summary":"L'autore kmeh ha scritto che il plugin di memoria Agent sul mercato presenta architetture difettose. Comunemente generano frammenti dai record di sessione e li memorizzano nel database RAG, iniettando i segmenti più simili ogni volta che le ricerche su prompt, ma affrontano problemi come la ricerca per similarità che non giudica la correttezza, frammenti che mancano di contesto e non auditabilità.","category":"行业动态","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"Gli Agent non richiedono plugin di memoria; richiedono memoria basata su documenti - Aioga Notizie IA","description":"L'autore kmeh ha scritto che il plugin di memoria Agent sul mercato presenta architetture difettose. Comunemente generano frammenti dai record di sessione e li memorizzano nel data...","url":"https://www.aioga.com/it/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:03:49.130Z"},"nl":{"title":"Agents hebben geen geheugen-plugins nodig; ze hebben documentgebaseerd geheugen nodig","summary":"Auteur kmeh schreef dat de Agent-geheugenplugin op de markt gebrekkige architecturen heeft. Ze genereren vaak fragmenten uit sessierecords en slaan deze op in de RAG-database, waarbij ze bij elke promptzoekopdracht de meest vergelijkbare segmenten injecteren, maar ondervinden problemen zoals het niet beoordelen van correctheid door de similariteitszoek, fragmenten die context missen en niet-auditeerbaarheid.","category":"行业动态","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"Agents hebben geen geheugen-plugins nodig; ze hebben documentgebaseerd geheugen nodig - Aioga AI-nieuws","description":"Auteur kmeh schreef dat de Agent-geheugenplugin op de markt gebrekkige architecturen heeft. Ze genereren vaak fragmenten uit sessierecords en slaan deze op in de RAG-database, waar...","url":"https://www.aioga.com/nl/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:03:49.281Z"},"tr":{"title":"Ajanlar bellek eklentilerine ihtiyaç duymaz; belge tabanlı belleğe ihtiyaç duyarlar","summary":"Yazar kmeh, piyasadaki Agent bellek eklentisinin kusurlu mimarilere sahip olduğunu yazdı. Genellikle oturum kayıtlarından parçalar üretir ve RAG veritabanında saklarlar, her çağrı aramasında en benzer segmentleri enjekte ederler, ancak benzerlik aramalarının doğruluğu değerlendirememesi, parçaların bağlamdan kaçması ve denetlenemezlik gibi sorunlarla karşılaşırlar.","category":"行业动态","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"Ajanlar bellek eklentilerine ihtiyaç duymaz; belge tabanlı belleğe ihtiyaç duyarlar - Aioga AI Haberleri","description":"Yazar kmeh, piyasadaki Agent bellek eklentisinin kusurlu mimarilere sahip olduğunu yazdı. Genellikle oturum kayıtlarından parçalar üretir ve RAG veritabanında saklarlar, her çağrı...","url":"https://www.aioga.com/tr/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:04:13.336Z"},"vi":{"title":"Agent không cần plugin bộ nhớ; chúng cần bộ nhớ dựa trên tài liệu","summary":"Tác giả kmeh viết rằng plugin bộ nhớ Agent trên thị trường có kiến trúc không hoàn hảo. Chúng thường tạo ra các đoạn từ các bản ghi phiên và lưu trữ chúng trong cơ sở dữ liệu RAG, mỗi lần tìm kiếm nhắc nhở lại các đoạn giống nhất, nhưng lại gặp các vấn đề như tìm kiếm tương đồng không đánh giá được độ chính xác, các đoạn thiếu ngữ cảnh và không thể kiểm tra được.","category":"行业动态","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"Agent không cần plugin bộ nhớ; chúng cần bộ nhớ dựa trên tài liệu - Tin tức AI Aioga","description":"Tác giả kmeh viết rằng plugin bộ nhớ Agent trên thị trường có kiến trúc không hoàn hảo. Chúng thường tạo ra các đoạn từ các bản ghi phiên và lưu trữ chúng trong cơ sở dữ liệu RAG,...","url":"https://www.aioga.com/vi/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:04:13.444Z"},"id":{"title":"Agen tidak memerlukan plugin memori; mereka membutuhkan memori berbasis dokumen","summary":"Penulis kmeh menulis bahwa plugin memori Agen di pasaran memiliki arsitektur yang cacat. Mereka biasanya menghasilkan fragmen dari catatan sesi dan menyimpannya di database RAG, menyuntikkan segmen paling mirip setiap kali pencarian prompt, tetapi menghadapi masalah seperti pencarian kesamaan yang gagal menilai kebenaran, fragmen yang hilang konteks, dan ketidakterauditabilitas.","category":"行业动态","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"Agen tidak memerlukan plugin memori; mereka membutuhkan memori berbasis dokumen - Berita AI Aioga","description":"Penulis kmeh menulis bahwa plugin memori Agen di pasaran memiliki arsitektur yang cacat. Mereka biasanya menghasilkan fragmen dari catatan sesi dan menyimpannya di database RAG, me...","url":"https://www.aioga.com/id/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:04:13.369Z"},"th":{"title":"เอเจนต์ไม่จําเป็นต้องใช้ปลั๊กอินหน่วยความจํา; ต้องการหน่วยความจําแบบเอกสาร","summary":"ผู้เขียน kmeh เขียนว่าปลั๊กอินหน่วยความจํา Agent ในตลาดมีสถาปัตยกรรมที่มีข้อบกพร่องโดยปกติจะสร้างชิ้นส่วนจากระเบียนเซสชันและเก็บไว้ในฐานข้อมูล RAG โดยจะแทรกชิ้นส่วนที่คล้ายกันมากที่สุดทุกครั้งที่มีการค้นหา prompt แต่พบปัญหา เช่น การค้นหาความคล้ายคลึงกันไม่สามารถตัดสินความถูกต้องได้ ชิ้นส่วนข้อมูลขาดบริบท และไม่สามารถตรวจสอบได้","category":"行业动态","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"เอเจนต์ไม่จําเป็นต้องใช้ปลั๊กอินหน่วยความจํา; ต้องการหน่วยความจําแบบเอกสาร - ข่าว AI Aioga","description":"ผู้เขียน kmeh เขียนว่าปลั๊กอินหน่วยความจํา Agent ในตลาดมีสถาปัตยกรรมที่มีข้อบกพร่องโดยปกติจะสร้างชิ้นส่วนจากระเบียนเซสชันและเก็บไว้ในฐานข้อมูล RAG โดยจะแทรกชิ้นส่วนที่คล้ายกันมากที...","url":"https://www.aioga.com/th/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:04:38.063Z"},"pl":{"title":"Agenci nie wymagają wtyczek pamięci; wymagają pamięci opartej na dokumentach","summary":"Autor kmeh napisał, że wtyczka pamięci Agent dostępna na rynku ma wadliwą architekturę. Często generuje fragmenty z rekordów sesji i przechowuje je w bazie danych RAG, wstrzykując najbardziej podobne segmenty przy każdym przeszukiwaniu promptu, ale napotykają takie problemy jak wyszukiwanie podobieństw nie oceniające poprawności, fragmenty brakujące kontekstu oraz brak audytu.","category":"行业动态","source":"Hacker News 热门","aggregationSource":"Hacker News 热门","pageTitle":"Agenci nie wymagają wtyczek pamięci; wymagają pamięci opartej na dokumentach - Aioga Wiadomości AI","description":"Autor kmeh napisał, że wtyczka pamięci Agent dostępna na rynku ma wadliwą architekturę. Często generuje fragmenty z rekordów sesji i przechowuje je w bazie danych RAG, wstrzykując...","url":"https://www.aioga.com/pl/news/oactlhauxke4xq1kl9xt97qol/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"f56709246462da08","translatedAt":"2026-10-04T12:04:36.890Z"}},"evidenceTier":"verified-news","reviewStatus":"automated-ingest","indexable":true,"editorialCover":""}}