跳到主内容
@wquguru
精选70MKT1(Emily Kramer · RSS)产品与增长

24个MCP工作流:将GTM技术栈接入Claude

24 MCP Workflows to Bring Your GTM Stack into Claude

原文
发到 X

👋 This is a bi-monthly free edition of MKT1 Newsletter brought to you by our MKT1 MCP Showcase partners: Zapier, Framer, Softr, Attio, Airtable, Profound & Mutiny.

👋 这是MKT1新闻通讯的双月刊免费版,由我们的MKT1 MCP展示合作伙伴带来:Zapier、Framer、Softr、Attio、Airtable、Profound和Mutiny。

🔌 MKT1-ify your LLM with the MKT1 MCP: Bring ~40 skills we built for B2B Startup marketers to Claude Chat, Cowork, or Code. Add it to Claude in 1 click →

🔌 使用MKT1 MCP将您的LLM进行MKT1化:将我们为B2B初创公司营销人员构建的约40项技能带到Claude Chat、Cowork或Code中。一键添加到Claude →

⬆️ Upgrade to a paid MKT1 subscription for: Full access to our MKT1 MCP | 100+ templates & resources | Post to MKT1 Job Board | MKT1 Newsletter Archive | $40K+ in discounts in the MKT1 Perk Stack

⬆️ 升级到付费MKT1订阅,即可获得:完全访问我们的MKT1 MCP | 100多个模板和资源 | 发布到MKT1招聘板 | MKT1新闻通讯存档 | MKT1福利包中价值超过4万美元的折扣

Subscribe now

立即订阅

We’ve entered the MCP era of AI at work, and there’s no turning back. You can’t build a truly AI-native system for work without connecting it to the rest of your stack.

我们已经进入了工作中AI的MCP时代,而且没有回头路。如果不将AI连接到您技术栈的其他部分,就无法构建真正AI原生的系统。

Why? While Claude’s pretty useful on its own, when you connect it to other tools in your GTM stack it becomes infinitely more useful. And every GTM product worth its salt already has a robust MCP or CLI (command-line interface).

为什么?虽然Claude本身相当有用,但当您将其连接到GTM技术栈中的其他工具时,它会变得无限有用。而且每个名副其实的GTM产品都已经拥有强大的MCP或CLI(命令行界面)。

All of this can be overwhelming. Many marketers are still getting comfortable breaking out of ChatGPT or Claude Chat and moving into Cowork, Code, or Codex, so adding connectors is a whole other can of worms. Whether you’ve never used an MCP, have 20 connected, or are running MCPs and CLIs in Claude Code all day, it can be hard to know if you’re doing it right. Don’t worry, I’ll define all these acronyms and terms in the MCP 201 section below.

所有这些都可能让人不知所措。许多营销人员仍在适应从ChatGPT或Claude Chat转向Cowork、Code或Codex,因此添加连接器完全是另一回事。无论您从未使用过MCP、已经连接了20个,还是整天在Claude Code中运行MCP和CLI,都可能很难知道自己的做法是否正确。别担心,我将在下面的MCP 201部分定义所有这些缩写和术语。

If you've been in this marketing-startup game for a while, you'll remember when “platform” and “integrations” swept SaaS tools. That's now happening with AI, and MCPs are the mechanism. We’ve seen this before!

如果您在这个营销初创公司领域待过一段时间,您会记得“平台”和“集成”席卷SaaS工具的时代。现在这种情况正在AI领域发生,而MCP就是其机制。我们以前见过这种情况!

With this in mind, we hosted an MKT1 MCP Showcase on 7/29, with 7 GTM tools live demoing their MCPs in Claude. The goal was to skip the fluff and demo workflows you can try right now.

考虑到这一点,我们在7月29日举办了一场MKT1 MCP展示会,有7个GTM工具在Claude中现场演示了它们的MCP。目标是跳过花哨的内容,直接演示您现在就可以尝试的工作流程。

So let's get to it. I'll break down what you need to know to get the most out of MCPs, define the other ways to connect your tools, and then get into the workflows from our 7 GTM tools (including video replays from the event to bring it all to life).

那么让我们开始吧。我将分解您需要了解的内容,以充分利用MCP,定义连接工具的其他方式,然后进入我们7个GTM工具的工作流程(包括活动视频回放,让一切生动起来)。

In this newsletter:

本期新闻通讯内容包括:

  • MCP 201: Ways to connect Claude to other tools, get more out of your MCPs, and solve your most common LLM complaints.
  • 21 workflows from 7 GTM tools: How to implement each one, plus video replays of the live demos from our MCP Showcase
  • Featuring: Zapier | Framer | Softr | Attio | Airtable | Profound | Mutiny
  • 3 workflows using the MKT1 MCP: Add MKT1's B2B marketing skills to your Claude workflows, from hiring and campaign briefs to auditing your finished work
  • Watch the full video replay here →
  • Some “inside baseball” for my most loyal MKT1 readers: Last edition I kicked off a 3-part series on multiplayer Claude. This newsletter isn’t part 2, it’s more of a first cousin to that series, as you can’t build a multiplayer system without MCPs. Part 2 will be out in 2 weeks though!
  • MCP 201:将Claude连接到其他工具的方法,从MCP中获得更多价值,并解决您最常见的LLM抱怨。
  • 来自7个GTM工具的21个工作流程:如何实施每个工作流程,以及我们MCP展示会现场演示的视频回放
  • 特色工具:Zapier | Framer | Softr | Attio | Airtable | Profound | Mutiny
  • 使用MKT1 MCP的3个工作流程:将MKT1的B2B营销技能添加到您的Claude工作流程中,从招聘和活动简报,到审计您完成的工作
  • 在此观看完整视频回放 →
  • 一些“内部消息”给我最忠实的MKT1读者:上一期我开启了关于多人协作Claude的三部分系列。这期通讯不是第二部分,更像是该系列的近亲,因为没有MCP你就无法构建多人协作系统。不过第二部分将在两周后发布!

Subscribe now

立即订阅

Watch the full replay

观看完整回放

We’ve also broken down the replay into chapters for each presenter throughout the newsletter if you’d prefer to watch that way.

我们还把回放按每位演讲者分成了章节,如果你喜欢这种方式观看,可以在通讯中找到。

Thanks to our Showcase partners

感谢我们的展示合作伙伴

Here's who presented and what they demoed:

以下是演讲者及其演示内容:

Zapier MCP: Run actions across any app you connect in Zapier, including tools with no MCP of their own. Learn more →

Zapier MCP:在Zapier中连接的任何应用上执行操作,包括那些没有自己MCP的工具。了解更多 →

Framer CLI: Edit and publish your Framer site from Claude.

Framer CLI:从Claude编辑和发布你的Framer网站。

Offer: 30% off annual Pro with the code MKT1. Learn more →

优惠:使用代码MKT1享受年度Pro版30%折扣。了解更多 →

Softr MCP: Create and update Softr databases and live apps from Claude. Learn more →

Softr MCP:从Claude创建和更新Softr数据库和实时应用。了解更多 →

Offer: 2 free months of the Business plan for new users with the code MKT1-2MO.

优惠:新用户使用代码MKT1-2MO可免费获得2个月商业计划。

Attio MCP: Query your CRM records, run reports on deal data, and mine insights from calls, all from Claude. Learn more →

Attio MCP:从Claude查询CRM记录、运行交易数据报告,并从通话中挖掘洞察。了解更多 →

Offer: 15% off your first year, using this link.

优惠:使用此链接首年享受15%折扣。

Airtable MCP: Bring your bases into Claude to query records, build tables and interfaces, and automate processes. Learn more →

Airtable MCP:将你的base带入Claude,查询记录、构建表格和界面,并自动化流程。了解更多 →

Profound MCP: Analyze your AI search data and run Profound Agents from inside Claude. Free AEO report for MKT1 readers. Learn more →

Profound MCP:在Claude内分析你的AI搜索数据并运行Profound Agents。为MKT1读者提供免费AEO报告。了解更多 →

Mutiny MCP: Build branded landing pages, deal rooms, and decks from Claude. Learn more →

Mutiny MCP:从Claude构建品牌落地页、交易室和演示文稿。了解更多 →

The companies in this showcase and newsletter were hand-selected by me. I use their MCPs and think they’re valuable to my audience of marketers. In full disclosure, they also paid to participate (which makes events like this possible).

本次展示和通讯中的公司均由我亲自挑选。我使用他们的MCP,并认为它们对我的营销受众有价值。完全披露,他们也付费参与(这使得这样的活动成为可能)。

MCP 201: How to get the most out of Connectors

MCP 201:如何充分利用连接器

Connecting an MCP is just the start. Once connected, you can embed the functionality of these tools into everything you’re already doing in your LLM or AI harness. Before we jump into the workflows, I’ll explain how all of this works so they feel easier to understand and implement.

连接MCP只是开始。连接后,你可以将这些工具的功能嵌入到你在LLM或AI工具中已经做的一切中。在我们深入工作流程之前,我会解释这一切是如何运作的,以便它们更容易理解和实施。

If you're already comfortable with MCPs, skip on down to the first set of workflows, from Zapier.

如果你已经熟悉MCP,可以直接跳到第一组工作流程,来自Zapier。

1. Ways to connect to Claude to other tools

1. 将Claude连接到其他工具的方式

Let's get some definitions out of the way first:

让我们先明确一些定义:

  • MCP (aka MCP Server aka Model Context Protocol Server aka Claude Connector): How a tool tells Claude what it can do, so Claude can use that tool on your behalf. An MCP is essentially a bundle of actions, like send a Slack Message or pull AEO data from Profound.
  • AI Harness (Claude Cowork, Claude Code, or Codex): Think of a harness as an infrastructure layer that wraps an AI model and gives it the ability to interact with other things outside the LLM.
  • How it fits together: The MCP tells Claude what a tool can do, and the harness is what goes and does it.
  • MCP(又名MCP服务器,又名模型上下文协议服务器,又名Claude连接器):工具告诉Claude它能做什么,以便Claude代表你使用该工具。MCP本质上是一组操作,比如发送Slack消息或从Profound提取AEO数据。
  • AI 工具链(Claude Cowork、Claude Code 或 Codex):可以把工具链想象成一个基础设施层,它包裹着 AI 模型,使其能够与 LLM 之外的其他事物进行交互。
  • 它们如何配合:MCP 告诉 Claude 某个工具能做什么,而工具链则负责实际执行。

Personal aside: My brother and I never had much in common career-wise. He builds ropes courses, using a harness. And now I build with AI, using a harness. Same same. Now strap in (harness pun!), let’s get back to the real content.

个人闲话:我和我兄弟在职业上从来没什么共同点。他建造绳索课程,使用安全带。现在我用 AI 构建,也使用工具链。一样一样。现在系好安全带(工具链的双关语!),让我们回到正题。

No MCP, no problem!

没有 MCP,也没问题!

There are other ways to bring tools into your AI system, and sometimes they give you even more functionality than an MCP would.

还有其他方法可以将工具引入你的 AI 系统,有时它们甚至能提供比 MCP 更多的功能。

Once you have an AI harness and/or are willing to run a few slightly more advanced prompts, you can connect to tools that don’t have an MCP yet or go beyond what their MCP supports.

一旦你有了 AI 工具链,并且/或者愿意运行一些稍微高级的提示词,你就可以连接到那些还没有 MCP 的工具,或者超越其 MCP 所支持的范围。

Note: For the rest of this newsletter, I’ll use Claude Code as my default AI harness for simplicity, but you can do much of this in other AI harnesses, from Claude Cowork to Codex.

注意:在本通讯的其余部分,为简单起见,我将使用 Claude Code 作为默认的 AI 工具链,但你在其他 AI 工具链中也可以完成大部分操作,从 Claude Cowork 到 Codex。

Here are the main ways to connect your tools to Claude:

以下是将你的工具连接到 Claude 的主要方法:

2. How to get more out of your MCPs

2. 如何更好地利用你的 MCP

Skills + MCPs + routines = magic in Claude. Well, most of the time.

技能 + MCP + 例程 = Claude 中的魔法。嗯,大多数时候是这样。

You may have connected a bunch of MCPs already (though be careful how many, since they can slow things down). But a lot of people stop there and never really incorporate them into their workflows.

你可能已经连接了很多 MCP(但要小心数量,因为它们可能会拖慢速度)。但很多人止步于此,从未真正将它们融入工作流程。

3 concepts will help you get much more out of your connectors: using multiple MCPs in the same session, wrapping a skill around an MCP so it works your way every time, and putting workflows on a schedule so they run without you.

3 个概念将帮助你更好地利用你的连接器:在同一会话中使用多个 MCP,围绕 MCP 包装技能以便每次都按你的方式工作,以及将工作流程安排到日程中以便自动运行。

Once you get a handle (or a harness?!) on all of this, it can genuinely change how you work. Claude starts to become a command center for your whole workday.

一旦你掌握了所有这些(或者说工具链?!),它就能真正改变你的工作方式。Claude 开始成为你整个工作日的指挥中心。

Here’s how each of these 3 concepts works:

以下是这 3 个概念各自的工作原理:

3. How MCPs solve your most common LLM complaints

3. MCP 如何解决你对 LLM 最常见的抱怨

When helping each company choose workflows to demo, and then watching all 7 demos live, the same handful of patterns kept coming up. MCPs and CLIs solve many of the limitations you run into when using an LLM on its own.

在帮助每家公司选择要演示的工作流程,并现场观看全部 7 个演示时,同样的几个模式反复出现。MCP 和 CLI 解决了许多你在单独使用 LLM 时遇到的限制。

Here’s how connectors help at a high level, across workflows:

以下是连接器在高层面上如何跨工作流程提供帮助:

Avoiding slop

避免劣质输出

  • The problem: Working from a prompt alone, Claude produces generic output. It doesn't know your product, your customers, or your standards, so what comes back could have been made for any company.
  • How MCPs solve it: By connecting Claude to the context that already lives in your tools (e.g. positioning guides in a doc, design guidelines in Figma), it can pull that knowledge directly into your sessions, workflows, and skills instead of relying on the prompt alone.
  • Examples: When Claude has your web design guidelines from Framer, it can move much faster making new pages.
  • 问题:仅凭提示词,Claude 会产生通用输出。它不了解你的产品、你的客户或你的标准,因此返回的内容可能适用于任何公司。
  • MCP如何解决这个问题:通过将Claude连接到您工具中已有的上下文(例如文档中的定位指南、Figma中的设计规范),它可以直接将这些知识引入您的会话、工作流程和技能中,而不是仅仅依赖提示词。
  • 示例:当Claude拥有您在Framer中的网页设计指南时,它可以更快地创建新页面。

Adding a source of truth

添加事实来源

There are two distinct use cases here worth separating:

这里有两个不同的用例值得区分:

Consolidating reporting from many sources

整合来自多个来源的报告

更进一步:量化金融体系

看懂新闻只是起点——沿量化金融路径,把它变成能交付的工程能力

进入量化体系 →

相似阅读

另一事件,读法相近