推文
760 tweets · 188 sources账号:全部mattpocockukdair_aisimonwbchernyadocompletetrq212dexhorthyyetonefelixrieseberg0xblacklightdani_avila7zarazhangruiAlchainHustbadlogicgamesdoteypetergyangvikingmuteryancarsonClaudeDevskunchenguidleon7haoctatedeviamzhihuikieranklaassenmckaywrigleylennysanthdxrRhysSullivangarrytanjakevin7johnlindquistkarpathyservasyy_aitheo0xMovez9hillsRLanceMartinaidenybaidingyihylarucodermitchellhmitsuhikomvanhornomarsar0steipetezeegGeminiAppGergelyOroszZaynHaoaddyosmaniantirezdanshipperdillon_mulroyelonmuskewind_devidoubiccnummanalirealWeZZardswyxthsottiauxyiliush0xPaulius0x_rodyAYi_AInotesAnthropicAIBarret_ChinaBenJames_____CMGS1988ChadMoranDanielMiesslerDaveJDavidKPianoDimillianFactoryAIFardeemMFarzaTVGoogleDeepMindHamelHusainHiTw93HilaShmuelIfanJewIndieDevHaileyJack_W_LindseyJackywineJasonZXJiaxi_CuiJinjingLiangKhalidWarsaLinearUncleLinghuaJMatthewBermanMaxForAIMeari_V2_0_GMnilaxQingQ77RaillyHugoSaccc_cTaniyatweets_VincentLogicai_explorer25aibuilderclub_alexalbert__alliekmillerandrewfarahantoinecojpanxue201arkuy99artmanasynkimobbssppllvvbearliubentossellbridgemindaibuaaxhmcailynyongyongcgtwtscharmaine_kleechrisbarberchrisparkXclairevodelba_oliveirademishassabisdoodlesteindriaforallelithrarelvissunericzakariassonfrancoisfleuretgabriell_labiamsahaj_xyzimdigitalashishimvihvjarredsumnerjerryjliu0jianshuojiayuan_jyjlongsterjonas_nellejoshalbrechtjpschroederkaiofreitaskepanolexruslogancyangmattypnabeelnifinetninthbit_aioran_gepatrickcpejmanjohnpetradonkaprathamgrvprathyvshquanruzhuoxiuquant_sheepraroquerealCaigurepsiacerohanpaul_aisamasamuelstroscheisawyerhoodshadcnshao__mengshaogefenhaoshreyansjtechgirl1908techwith_ramthorstenballtianyitobiaswuptoddsaunderstricaltturingoututuretomuniswap12untraceable_thevanillaCitronvelvet_sharkwengtianxinwquguruxiaogaifunxicilionycombinatorzachlloydtweetsziwenxu_zodchiii
@adocomplete·6月18日
New on the Claude Community page: share what you've built with Claude and get featured.
Also learn about upcoming events, the ambassador program, and more! https://t.co/pQhyX338KC
Claude 社区页面新上线：分享你用 Claude 构建的作品并获得推荐展示，还可了解即将举办的活动、大使计划等。
@adocomplete·6月17日
This is amazing.
As someone that's not a game dev but has tried with modern AI tooling and not gotten far I am super excited to play with this!
@UnrealEngine
Unreal Engine 5.8 ships today with experimental MCP server support: Your sources, your pipeline and your workflow—simply configure the MCP plugin and connect to any agent. Get familiar with the MCP server and the PCG Primitive Plugin today and see what teams can build together: https://t.co/cDITLWWv2F
这太惊人了。 作为一个不是游戏开发者、但用现代 AI 工具尝试过却没走多远的人，我对这个超级兴奋！ 【引用 @UnrealEngine】Unreal Engine 5.8 今日发布，带实验性 MCP server 支持： 你的素材、你的管线、你的工作流——只需配置 MCP 插件即可接入任何 agent。现在就来了解 MCP server 和 PCG Primitive Plugin，看看团队能一起构建什么。
@adocomplete·6月12日
Keep building!
@claudeai
Claude Fable 5 has been out for a couple of days. Some projects people have already built with it:
继续搭建！ [引用 @claudeai]: Claude Fable 5 已发布几天了。来看看人们已经用它构建的一些项目：
@adocomplete·6月11日
Introducing Claude Corps
We’ll teach 1,000 fellows how to use Claude well, match them with nonprofits across America, and pay them to spend a year—full-time, in-person—helping host organizations to advance their missions. https://t.co/kgsJmljOns
介绍 Claude Corps 计划 我们将培训 1,000 名 fellow 学员如何善用 Claude，将他们与全美各地非营利组织配对，并资助他们花一整年时间——全职、驻场——帮助合作机构推进使命。
@adocomplete·6月10日
One other thing I'll say about Fable 5 is that it works so well with @Cloudflare ecosystem. Building and deploying full stack apps at internet scale has never been easier.
Wrangler 🤝 Claude Fable 5 ftw! https://t.co/alKDmsLTdu
Fable 5 和 Cloudflare 生态配合得非常好。在互联网规模上构建和部署全栈 app 从未如此简单。 Wrangler 🤝 Claude Fable 5，绝了！
@adocomplete·6月10日
Fable 5 crushes at building iOS apps. Foundation Models support makes it even easier to add AI to your iOS apps.
@ClaudeDevs
New for Apple developers: Foundation Models support for Claude lets developers use Apple's Foundation Models framework to call Claude for multi-step reasoning, code generation, and longer context. https://t.co/7EbHlWBBSe
Fable 5 做 iOS app 真的是降维打击。Foundation Models 支持让在 iOS app 里接入 AI 变得更简单了。 [引用 @ClaudeDevs]: Apple 开发者新功能：Claude 的 Foundation Models 支持让开发者可以用 Apple 的 Foundation Models framework 调用 Claude，实现多步推理、代码生成和更长 context 处理。
@adocomplete·6月10日
Long running tasks with Fable 5 is no joke.
Combined with ultracode and /remote-control, it really feels like you have access to raw intelligence in the palm of your hands.
Fable 5 的长任务能力真不是开玩笑。配合 ultracode 和 /remote-control，真的感觉把原始智能握在手心里。
@adocomplete·6月9日
Introducing Claude Fable 5, our most capable public model ever.
Best-in-class for software engineering, scientific research, knowledge work, and vision.
Available today on all paid plans, in Claude Code, on the Claude API, and all major cloud platforms. https://t.co/hEhR5eUxgA
隆重推出 Claude Fable 5，我们迄今最强的公开模型。 软件工程、科学研究、知识工作和视觉理解均达到业界最佳水准。 今日起在所有付费计划、Claude Code、Claude API 及各大云平台正式上线。
@adocomplete·6月4日
8x as much code written vs baseline. 😱
@AnthropicAI
Today, Anthropic engineers on average ship 8x as much code per quarter as they did compared to 2021-2025. https://t.co/QCc9cqGgf4
代码产出量是基准的 8 倍 😱 （引用 @AnthropicAI：今天，Anthropic 工程师平均每季度的代码产出量是 2021-2025 年时期的 8 倍。）
@adocomplete·6月3日
oh no...or oh yes? https://t.co/kHnSAoyMzV
@adocomplete
Gave Claude an open-ended overnight /goal. Can't wait to see what it builds while I sleep. 🤞laptop doesn't try to update or something interrupts it.
哦不……还是哦耶？[链接] [引用 @adocomplete]: 给 Claude 布置了一个开放式的过夜 /goal 任务。迫不及待想看它睡觉期间能搭出什么。🤞 希望笔记本电脑别试图更新或被什么打断。
@adocomplete·6月2日
Workflows are a pretty big deal. Learn why.
@trq212
Workflows are the biggest upgrade to Claude Code’s capabilities since skills and subagents. I dove deep into it with @sidbid to figure out best practices, examples and more. I’m particularly excited about the non-technical tasks it enables for Claude Code.
Workflows 是相当重要的事。了解一下为什么。 [引用 @trq212]：Workflows 是继 skills 和 subagents 之后 Claude Code 能力的最大升级。 我和 @sidbid 深入研究了最佳实践和示例。我特别期待它为 Claude Code 开启的非技术类任务场景。
@adocomplete·6月1日
Rate limit resets! Go forth and code!
@ClaudeDevs
We've reset 5-hour and weekly rate limits for all users on Pro and Max plans. We fixed an issue that caused some Claude Code sessions to spawn excessive parallel subagents, burning through usage faster than expected.
@ClaudeDevs：我们已重置 Pro 和 Max 计划所有用户的 5 小时和每周 rate limit。 我们修复了一个问题：该问题导致部分 Claude Code session 产生过多并行 subagent，导致用量消耗速度远超预期。 --- @adocomplete：Rate limit 重置了！去尽情写代码吧！
@adocomplete·5月29日
Protip if you use Claude for Chrome a ton: Install Chrome Beta to use as your human daily driver. It's super stable, and you can continue to do human browsing while Claude can have uninterrupted access to regular Chrome.
如果你大量使用 Claude for Chrome，小技巧：安装 Chrome Beta 作为你自己的日常浏览器。它非常稳定，你可以继续用它正常上网，而 Claude 则独占常规 Chrome，互不干扰。
@adocomplete·5月28日
Opus 4.8 is awesome but that's expected. The unsung hero of this release for me is dynamic workflows.
Claude plans your task, fans it out to tens or hundreds of parallel subagents, verifies their work, and iterates until the results converge into one coordinated answer. https://t.co/0hygzWB2mH
Opus 4.8 很棒，这在意料之中。对我来说，这次发布真正被低估的亮点是 dynamic workflows。 Claude 规划你的任务，将其分发给数十甚至数百个并行 subagent，验证它们的工作，反复迭代，直到结果收敛成一个协调一致的答案。
@adocomplete·5月28日
Opus 4.8 is here!
Sharper judgment. More honest. Works longer. Same price.
Plus:
* Fast mode at 2.5x speed and 3x cheaper than before
* Dynamic workflows run 100s of parallel subagents in Claude Code
* Effort control in Claude app
Ready for action wherever you get your Claude
Opus 4.8 来了！ 判断力更敏锐，更诚实，工作时间更长，价格不变。 另外： * Fast mode 速度提升 2.5x，成本降至原来 1/3 * Claude Code 支持动态工作流，可并行运行数百个 subagent * Claude app 新增 effort 控制 随时随地，Claude 已就绪。
@adocomplete·5月21日
Any system offering comfort with no visible cost is the cost itself.
凡是看似零代价就能给你舒适感的系统，这个系统本身就是那个代价。
@adocomplete·5月20日
So many must know pro-tips on how to get the most out of computer and browser use with Claude.
If you've used computer use and had issues getting Claude to click the right thing (like I have), I'm confident this post will have at least 2 things to help you. https://t.co/TzsM9cbSc1
关于如何用好 Claude computer use 和 browser use，有太多必知的实用技巧了。 如果你用过 computer use 却遇到 Claude 点错位置的问题（我也遇到过），我相信这篇文章至少能给你 2 个有用的解法。
@adocomplete·5月19日
Congrats on the launch. Looks super promising!
@OfficialLoganK
Welcome to Gemini 3.5 Flash, our most powerful model to date. It pushes the frontier of intelligence, speed, and cost putting 3.5 Flash in a class of its own. We spent the last 6 months making sure Flash is great for real world use cases. It's available everywhere now! https://t.co/03QG3fSh9b
恭喜发布，看起来非常有潜力！ 【引用 @OfficialLoganK】：欢迎来到 Gemini 3.5 Flash——我们迄今最强大的模型。它在智能、速度和成本上全面突破，将 3.5 Flash 推向独一档的水平。 过去 6 个月我们专注于让 Flash 在真实世界场景中表现优秀。现已全面上线！
@adocomplete·5月19日
Self-hosted sandboxes and MCP tunnels in Claude Managed Agents!
@claudeai
Live from Code with Claude London: we're launching self-hosted sandboxes (public beta) and MCP tunnels (research preview) in Claude Managed Agents. Run agents inside your own perimeter, with your security controls applied by default. https://t.co/cxvmk3feHp
Claude Managed Agents 中上线了自托管沙箱和 MCP tunnels！ [引用 @claudeai]: 直播来自 Code with Claude London：我们正在 Claude Managed Agents 中发布自托管沙箱（公测）和 MCP tunnels（研究预览版）。 在你自己的安全边界内运行 agent，默认应用你的安全控制。
@adocomplete·5月15日
5-hour and weekly rate limits have been reset for Claude Code.
Go building something fun this weekend :D
@ClaudeDevs
Happy Friday! We've reset everyone's 5-hour and weekly rate limits.
Claude Code 的 5 小时和每周用量限制已重置。 这个周末去做点有趣的东西吧 :D 【引用 @ClaudeDevs】周五快乐！我们已为所有人重置了 5 小时和每周用量限制。
@adocomplete·5月13日
Claude Code weekly limits are increasing 50%, now through July 13, stacking with the 2x increase to 5-hour limits announced last week.
Live now for all Pro, Max, Team, and seat-based Enterprise users.
Go forth and build! https://t.co/2QWnQt7U2z
Claude Code 每周使用限额提升 50%，有效期至 7 月 13 日，与上周宣布的 5 小时限额 2 倍提升叠加生效。 现已对所有 Pro、Max、Team 和按席位计费的 Enterprise 用户生效。 开干吧！
@adocomplete·5月13日
An update on Claude Agent SDK usage with Claude Subscription plans.
@ClaudeDevs
Starting June 15, paid Claude plans can claim a dedicated monthly credit for programmatic usage. The credit covers usage of: - Claude Agent SDK - claude -p - Claude Code GitHub Actions - Third-party apps built on the Agent SDK
Claude Agent SDK 在 Claude 订阅套餐中的使用更新。 [引用 @ClaudeDevs]：从 6 月 15 日起，付费 Claude 套餐可以领取专属的每月额度，用于程序化使用。 该额度覆盖： - Claude Agent SDK - claude -p - Claude Code GitHub Actions - 基于 Agent SDK 构建的第三方应用
@adocomplete·5月11日
So excited for Agent view (research preview) in Claude Code.
Run claude agents to list and start multiple sessions. Each one runs w/o taking up a terminal tab.
See what's running, what's blocked, and what's done. Reply inline or jump in and out of sessions with ease. https://t.co/zXlD9cEOgh
太期待 Claude Code 的 Agent view（研究预览版）了。 可以用 claude agents 命令列出并启动多个 session，每个都在后台运行，不占终端 tab。 一目了然地看到哪些在跑、哪些卡住了、哪些已完成。随时 inline 回复或自由切入/退出各个 session。
@adocomplete·5月11日
Claude Platform on AWS!!!
@claudeai
The Claude Platform on AWS is now generally available. AWS customers get the full set of Claude API features, with AWS authentication, billing, and commitment retirement. https://t.co/9WPrEgCpkg
Claude Platform 现已在 AWS 上正式 GA（全面可用）。 [引用 @claudeai]: AWS 用户现可获得完整的 Claude API 功能集，并支持 AWS 原生认证、账单计费和 commitment 消费抵扣。
@adocomplete·5月8日
Ocean -> rain -> tap water -> distilled water
The ocean is the pretraining corpus and we've just been finetuning for 4 billion years.
海洋 → 雨水 → 自来水 → 蒸馏水 海洋是预训练语料库，而我们只是在用 40 亿年做 fine-tuning。
@adocomplete·5月6日
New in Claude Managed Agents:
- Dreaming: reviews past sessions to improve agents over time
- Outcomes: set a quality rubric, agent iterates until it passes
- Multiagent Orchestration: parallel specialists
- Webhooks
Dreaming in research preview; rest in public beta.
Claude Managed Agents 新功能： - Dreaming：回顾历史 session，持续改进 agent 表现 - Outcomes：设定质量 rubric，agent 迭代直到通过 - Multiagent Orchestration：并行专家 agent - Webhooks Dreaming 进入研究预览，其余已公测。
@adocomplete·5月6日
We are increasing usage limits for Claude users so you can build more. https://t.co/ub1Yb8ouV2
我们正在为 Claude 用户提高使用限额，让你可以构建更多。
@adocomplete·4月9日
Pair Opus as an advisor with Sonnet as an executor in the API!
@claudeai
We're bringing the advisor strategy to the Claude Platform. Pair Opus as an advisor with Sonnet or Haiku as an executor, and get near Opus-level intelligence in your agents at a fraction of the cost. https://t.co/fRkegyMs5t
在 API 中将 Opus 作为顾问与 Sonnet 作为执行者配对！ [引用 @claudeai]: 我们正在将 advisor 策略带到 Claude Platform。 将 Opus 作为顾问，与 Sonnet 或 Haiku 作为执行者配对，以极低的成本在你的 agent 中获得接近 Opus 水平的智能。
@adocomplete·4月8日
Someone on Reddit taught Claude to talk like a caveman and reduce token usage by up to 75%.
Kevin Malone was so ahead of his time. Pure genius. 🤣 https://t.co/hDLpj7aung
有人在 Reddit 上教 Claude 像穴居人一样说话，结果 token 用量降低了最多 75%。 Kevin Malone 真的是时代先驱，纯粹的天才。🤣
@adocomplete·4月8日
Building, deploying, and scaling agents just became a whole lot easier with Claude Managed Agents.
You define your agent's tasks, tools, and guardrails. We run it on our infrastructure, so you can go from prototype to production to scale quickly and iterate even faster. https://t.co/6mAAiUXuvB
构建、部署、扩展 agent 现在容易多了——Claude Managed Agents 来了。 你只需定义 agent 的任务、工具和护栏，我们负责在自己的基础设施上运行，让你快速从原型走向生产，迭代更快。
@adocomplete·4月7日
Claude Mythos Preview is a general-purpose, unreleased frontier model that reveals a stark fact: AI models have reached a level of coding capability where they can surpass all but the most skilled humans at finding and exploiting software vulnerabilities.
@AnthropicAI
Introducing Project Glasswing: an urgent initiative to help secure the world’s most critical software. It’s powered by our newest frontier model, Claude Mythos Preview, which can find software vulnerabilities better than all but the most skilled humans. https://t.co/NQ7IfEtYk7
Claude Mythos Preview 是一个通用的、尚未发布的前沿模型，它揭示了一个严峻事实：AI 模型在编码能力上已经达到了一个新水平——在发现和利用软件漏洞方面，能够超越除最顶尖安全研究员以外的所有人类。 [引用 @AnthropicAI]：介绍 Project Glasswing：一项帮助保护全球最关键软件的紧急计划。 它由我们最新的前沿模型 Claude Mythos Preview 驱动，该模型发现软件漏洞的能力超越了除最顶尖安全研究员以外的所有人类。
@adocomplete·4月7日
I keep hearing "taste" is the only moat left. Hard disagree.
It's always framed as this abstract "you have it or you don't, you'll know it when you see it" thing which conveniently can't be taught, measured, or argued with.
I think the real moat has always and will always be deeply understanding what the user needs and providing solutions for it.
Building shallow-depth products is easy, even easier now with AI, but talking to customers, understanding their workflows, needs, wants, and working through the data to find innovations and building solutions for them will always be a winning strategy.
我一直听到「品味是唯一护城河」这个说法。强烈不同意。 这个概念总被包装成「要么有要么没有、见到就知道」的东西——恰好因此无法被教授、衡量或辩驳。 我认为真正的护城河从来都是，也永远会是：深度理解用户需要什么，并为此提供解决方案。 构建浅层产品很容易，有了 AI 更容易。但与客户深度对话、理解他们的工作流、需求、期望，从数据中发现创新点并构建解决方案——这永远是制胜策略。
@adocomplete·3月26日
👇
@trq212
To manage growing demand for Claude we're adjusting our 5 hour session limits for free/Pro/Max subs during peak hours. Your weekly limits remain unchanged. During weekdays between 5am–11am PT / 1pm–7pm GMT, you'll move through your 5-hour session limits faster than before.
👇 【引用 @trq212】为应对 Claude 日益增长的使用需求，我们将在高峰时段调整免费/Pro/Max 订阅用户的 5 小时会话限制。每周总限额不变。 工作日太平洋时间早 5 点至上午 11 点（北京时间晚 8 点至次日凌晨 2 点）期间，你的 5 小时会话限额消耗速度将比以前更快。
@adocomplete·3月25日
Claude Code never seizes to amaze me. Built this entire game in a single prompt. The exact prompt in an empty directory:
"can you build an isometric game using phaser.js that has turn based combat and you play against an AI. add levels, a story, and good UI. ultrathink." https://t.co/qikbMC42jF
Claude Code 从未停止让我惊叹。用一条 prompt 在空目录里构建了整个游戏。完整 prompt 如下： 「用 phaser.js 构建一个等距视角游戏，包含回合制战斗、玩家对抗 AI、关卡系统、故事剧情和优质 UI。ultrathink。」
@adocomplete·3月24日
Claude Code's new auto mode is here.
Skip the permission prompts. A classifier reviews every action before it runs. Safe = go. Risky = blocked.
Less friction, more control. Research preview on Team plan now, Enterprise & API rolling out soon. https://t.co/S5PxIQyAaG
Claude Code 的 auto 模式来了。 跳过权限确认弹窗。每个操作执行前由分类器审查：安全则放行，高风险则拦截。 更少摩擦，更强管控。现已在 Team 计划上开放 research preview，Enterprise 和 API 即将推出。
@adocomplete·3月23日
Now this is next level!
@claudeai
You can now enable Claude to use your computer to complete tasks. It opens your apps, navigates your browser, fills in spreadsheets—anything you'd do sitting at your desk. Research preview in Claude Cowork and Claude Code, macOS only. https://t.co/sVymgmtEMI
这真的是下一个层级！ 【转引 @claudeai】你现在可以让 Claude 控制你的电脑来完成任务了。 它可以打开应用、浏览网页、填写表格——任何你坐在电脑前会做的事情。 Research preview 已在 Claude Cowork 和 Claude Code 中上线，目前仅限 macOS。
760 tweets · 188 sources