--- author: How I AI date: '2026-05-07' guest: '' layout: post.njk source: https://www.youtube.com/watch?v=efVfydaUIrM speaker: How I AI tags: - agentic-workflow - developer-tools - automation - ai-agents title: Anthropic's Claude Code 重大更新：Agentic 工作流新纪元 summary: 本次更新概述了 Anthropic 在 'Code with Claude' 活动上发布的五项重要功能，旨在增强开发者使用 Claude Code 和 Claude API 的能力。主要包括：Claude Code 应用中的 'Routines' 功能，实现任务的自动化调度；Claude API 中的 'Outcomes'，允许代理根据预设的 Rubric 自我评估和迭代；'Multi-agent Framework'，支持构建包含多达25个代理的协作团队；'Dreams'，一种用于整合和管理代理长期记忆的机制；以及整体使用限额的提升，使开发者能更频繁、更深入地使用这些工具。这些更新共同指向 Anthropic 致力于成为领先的 Agentic 平台。 insight: '' draft: true series: '' category: ai-application area: tech-engineering project: [] people: [] companies_orgs: - Anthropic products_models: - Claude Code - Claude API - Claude managed agents media_books: [] status: evergreen --- ### Anthropic“Code with Claude”活动发布五大Agentic新能力 本次“Code with Claude”开发者活动聚焦于增强开发者在代码编写和 Agentic 系统构建方面的能力。Anthropic 发布了多项重要更新，旨在提供更强大、更灵活的工具集，涵盖任务自动化、代理智能迭代、多代理协作以及记忆管理等方面，并提升了整体使用额度，以支持更广泛和深入的 AI 应用开发。
Original English Welcome back to How I AI. I'm Claire Vo, product leader and AI obsessive here on a mission to help you build better with these new tools. Today I attended Code with Claude, Anthropic's first developer event, and they announced some things in Claude Code and Claude managed agents I think you want to know about. I'm going to walk you through five things that launched today, how they work, what they are, and what I might build with them. We're going to keep it under 10 minutes and this is going to be a quick preview of what you'll see in your new Claude Code and Claude API products experience.
### Claude Code 应用更新：引入“Routines”实现任务自动化调度 Claude Code 应用迎来重大更新，其中最受期待的是 **“Routines”** 功能。这项新能力允许用户设置事件或动作的自动触发，能够基于预设的**计划（cron）**、**HTTP 请求**或 **GitHub Webhook** 来执行。例如，用户可以配置一个每周自动生成邮件通讯的 Routine，根据 `docs/change-log.md` 文件内容，撰写一份面向客户的、突出新功能的周报。该功能支持在云端或本地机器上运行，并能集成已有的 Connectors（如 Slack、GitHub），极大地提升了开发工作流的自动化水平。
Original English Okay, the first thing that shipped at Code with Claude that I think you want to know about are some updates to the Claude Code app. And one of the updates that I know we've all been waiting for is routines, the ability to trigger events or actions on a schedule. We love it. You know I love Open Claude and what I love about Open Claude are the crons. And so now Claude Code has that built in right here in the app. All you have to do is click new routine. You can either run it locally or remote. I'm going to run it locally. And I'm going to say weekly newsletter. This is something that I haven't been doing. I'm sorry if you're a Chat Period customer. I know I haven't done my newsletter. And I'm going to say, "Look at our change log and draft a newsletter for us to send weekly." And then I'm going to go in here and just say, "We have a change log.md in the docs folder. Review it every Monday and write a customer-facing newsletter based on the best customer-facing features we shipped. Don't talk about behind-the-scenes things like tech debt or security unless they really impress customers." Okay, so I'm going to do that. I'm going to run it daily or no, sorry, I'm going to run it weekly on Mondays at 6:00 a.m. And I think that's all I need to do. Oh, I'm going to select my folder um where my project is and then I'm going to create create. And now Claude Code will run my newsletter draft cron every week and then I can come back in here, grab the HTML. If I were being really fancy, I would hook this up to my newsletter platform, I would hook it up to my Slack and ping us, but again, this is very useful to me. This is something that I used to kick off manually in Claude Code every Monday, and now I can do that here in Claude Code on a schedule. So, how does that work? There are three trigger types. You can trigger them on a cron, which is a schedule, um HTTP or a GitHub web hook. So, you can do sort of a normal web hook or a GitHub web hook. You can trigger these three ways. Scheduled, like I just did, off of a GitHub action or a general web hook. So, you can hook it up to other systems to kick off a routine. All the stuff in connectors come along. So, I have Slack connected, I have GitHub connected, so you can use those things as part of your routine. And it can run in the cloud or it can run on your laptop, like I showed. And this is an example of a use case where you could say, "Weekly, I want you to check every PRD modified this week and check if it matches our rubric and post a summary to the team channel." So, that's item one.
### Claude API 增强：引入“Outcomes”实现代理智能迭代 Anthropic 在 Claude API 中引入了 **“Outcomes”** 功能，作为对现有代理模型（如 Codex 的“goal”）的类似增强。该功能允许用户定义代理完成任务的“完成标准”——一个 **Rubric**（以 Markdown 文件形式上传或内嵌）。代理会根据此 Rubric 进行自我评估和迭代，最多可进行 20 次尝试，直到达到预期的“完成”状态。这一机制特别适用于需要严格遵循规范的场景，如生成一份“ship-ready”的 PRD（产品需求文档），代理能够反复迭代，确保其输出符合所有优先级和技术能力要求。
Original English The second one is in Claude manage agents in the API. If you haven't paid attention, opening I released something in Codex called goal. You can do {slash} goal in beta in Codex and it'll basically bang its head against the problem, do what's called a rough loop, against the problem until it actually hits the goal. Anthropic released something very similar in the Claude API called outcomes. You define what done looks like for an agent, it can self-grade and iterate until it gets there. There's a couple interesting things you need to know about how outcomes are defined. They all anchor on what's called a rubric. So, there's a markdown file that's uploaded either through the files API or in line, and it's going to tell your agent what success looks like. Then, there is a grader, and it can do up to 20 iterations on the task to get to the outcome that you're going for. I want to walk through this one in a very specific example to make this just a little bit more concrete for folks. So, so imagine that I want you to ship a ship-ready PRD, and I don't know if you can relate to this. Often, you go through feedback cycles. You have to check it against priorities. You have to check it against technical capabilities. Now, using a Claude managed agent, you could in theory write a rubric, which is what does a good ship-ready um PRD look like, and then the agent can just take your PRD or your idea and iterate over and over and over again until it's fixed. Of course, you could expect Of course, you could expose this to your customers in an app like I might do for Chat PRD. But again, this idea of outcome is define a rubric, give the agent the task, let it bang its head against that at least 20 times till it gets it right. I think this really interesting model for agenda products and something I suspect many of us will use.
### Claude API 增强：支持强大的“Multi-agent Framework” Claude API 现在支持一个强大的 **“Multi-agent Framework”**，允许开发者通过 API 定义一个包含最多 25 个代理的协作团队。该框架具有明确的**层级结构**，包括一个**编排者（Orchestrator）**和多个**委托者（Delegates）**。每个代理都可以拥有自己独特的**工具集**。这种能力使得构建复杂的 Agentic 应用成为可能，例如，可以创建一个 PRD 编排者，并配备策略代理（模拟 CPO 声音）、评论代理（找出 PRD 中的漏洞）和评审代理（可能访问 GitHub 以优化技术实现）等子代理。这种团队协作模式为 Agentic 产品提供了前所未有的灵活性。
Original English The second thing I really love is a multi-agent framework supported in Claude managed agent. So, now you can through the API explicitly define a multi-agent team that's going to work against the same container, the same file system, up to 25, which is kind of amazing. You can have a a orchestrator and then delegates, and so there's explicit hierarchy, and each agent can have its own tool set. I think this is really cool cuz now you're able to define not just individual agents, but teams of agents programmatically through the API. And so, the example I would give for something like Chat PRD is you could have a PRD orchestrator. This is sort of like the master agent that is intended to define and drive the work across the team. And then you could have three pieces or three sub agents. A strategy agent that reflects the CPO voice, the critic agent that's sort of like supposed to poke at the holes in the PRD. I like being the critic agent. And then end review that can maybe have access to something like GitHub to optimize the technical implementation of the PRD. And so, you can define this as you see over here in the API. You define an agent in the API, you give it um a orchestrator level set of tools, and then you can define the sub agents um in the in the API with their own set of tools. And then you could expose that, as you can see here on the right, as three agents all working in parallel against the same problem owned by the coordinator or orchestrator level. Again, I think this is an interesting enhancement on the primitives of agents that people are going to be using quite a bit.
### “Dreams”功能：Agentic 记忆管理新原语 Anthropic 推出了名为 **“Dreams”** 的新功能，专注于 Agentic 系统的**记忆管理**。它提供了一个原语（primitive），可以作用于一系列 Agent session，用于整合和提炼这些会话中的关键信息，并将其以 Markdown 文件形式写入磁盘，供代理未来参考。该功能比喻为人类的“做梦”，即回顾白天的经历并选择性地固化重要记忆。虽然目前处于研究预览阶段，但“Dreams”预示着 Agentic 平台将集成更智能、更主动的记忆生成和管理机制，并可能进一步发展出“遗忘”能力。
Original English Okay, the next one I really like. It is Dreams. So, this is all about agent memory. Just to make it simple for folks, memory is basically the idea of writing markdown files to the file system your agent uses that helps it do a better job the next time. It's not that fancy. Often those files have a date on them, but you don't really have to overthink it. But, creating those memories is a little hard. And often a lot of the harnesses right now write memory on a hook. They write that on an event. And so, what they do is like when you close a session, it writes memory or when something happens, it writes memory. Or like with Open Claw, you can explicitly tell it to write memory. But, what I like about dreams, which is a very funny brand for an agentic memory product, but we'll allow it, is it's a primitive to call against a list of agent sessions. So, let's say you've done 50 things with your agent, it's an explicit call to take those 50 sessions, review them, and then come up with important memories to write to disk. And as I'm saying this, I guess this is what we do when we dream. We go through our day, we review it silently, and then we decide what to commit to memory. I don't know. I don't know if this is the perfect metaphor, but it's the one we got, and it looks great on a on a branded website. This one's in research preview, so I don't think everybody has access to this through API. I certainly don't have access to it, so I'm looking forward to touching it. But, why I think this one's important to know is it just gives you a frame of reference for how Anthropic and these labs are thinking about the primitives again of agents and agent memory, and you can predict that some framework like this is going to be integrated into agentic platforms or agentic products, where on some action or some regular cadence, you're going to review past sessions, and you're going to explicitly write the right things to disk so they can be referred to moving forward. Side note, I think we think a lot about agent memory, but not a lot about agent forget getting, so I'm looking forward to like the purge version of this, which is dreams that tell you what to forget. I don't know if that's like trauma erasure or whatever, but I think there's something interesting here.
### 使用限制提升：扩大 Claude Code 与 API 的使用额度 此次更新还包括对 **Claude Code 和 Claude API 的使用限制进行显著提升**。具体而言： * **Claude Code** 的 Pro、Max、Team 和企业版 5 小时使用限制翻倍。 * Pro 和 Max 计划的**峰值时段限制**将被取消。 * API 中 **Opus 模型**的速率限制也将提高。 这些调整意味着开发者将能够更频繁、更深入地使用这些工具，从而加速开发进程和 AI 应用的探索。
Original English Okay, and then number five, the only announcement people really care about, which is usage limits are up. So, starting today, Claude Code's 5-hour limits are now doubled across Pro, Max, Team, and seat-based enterprise platforms. Peak hours are going away for Pro and Max plans, and the rate limits for Opus models in the API are going up. So, we can all use these products more.
### 总结与未来展望 总而言之，Anthropic 在“Code with Claude”活动上推出的一系列更新，包括自动化的“Routines”、智能迭代的“Outcomes”、强大的“Multi-agent Framework”以及 Agentic 记忆管理的原语“Dreams”，并结合使用限额的提升，共同构建了一个更完善、更易于使用的 Agentic 开发平台。这些实用且前瞻性的功能表明 Anthropic 致力于成为开发者构建下一代 AI 应用的首选平台。
Original English Again, what do we see today at Code with Claude? Lots of other stuff. They might put data centers in space. There was a partnership with SpaceX. All sorts of fun things are happening, but the things that I think you can use immediately routines for scheduling tasks in Claude code, so you can get things done either on a webhook or on a schedule. Outcomes, which is the ability to set a rubric and task and have an agent work against that task at least 20 times to nail the rubric. Multi-agent orchestration, which allows you in the API to define an orchestrator role and sub-agents, I think up to 25, to get work done from different points of view with different tools. Dreams, which are a way to consolidate agent memory over sessions, over time, and do that on demand. And then finally, we can all use more Claude code, which makes everyone happy. I do not know if these limit increases apply to Claude design. I suspect they do not, but I hope they will. I think they launched a lot of stuff that we can use practically right now in Claude code and gave me some good ideas for what the future of agentic products look like. None of this was mind-blowing, but I think it's all really practical, and it just gives you the idea that Anthropic is trying to be the agent platform of choice for builders. I look forward to seeing what you build with it, and I'm going to go play with the API. Bye. Thanks so much for watching. [music] If you enjoyed this show, please like and subscribe here on YouTube or even better, leave us a comment with your thoughts. You can also find this podcast on Apple Podcasts, Spotify, or your [music] favorite podcast app. Please consider leaving us a rating and review, which will help others find the show. You can see all our episodes and learn more about the show at howiai.com. See you [music] next time.