Four Connectors, Not Forty

Four Connectors, Not Forty

Claude's directory now holds hundreds of integrations. You need four. Here's the chain that separates a stack that works from a pile of connected apps, plus five ready-made stacks for the business you're actually running.

⏱ 18 min read ● Beginner

Section 01

The problem stopped being availability

Two years ago the question was whether Claude could reach your tools at all. That question is dead. Depending on who's counting and what they count, Claude's connector directory holds somewhere between 369 and 439 integrations across as many as 30 categories, and one well-regarded guide from June puts the number of polished, one-click directory apps at closer to 50. Those numbers disagree because they're measuring different things: the curated directory, the full browsable list including third-party servers, and the subset that's genuinely one click.

The disagreement is the point. Anthropic doesn't publish a headline number at all, so every figure you'll see quoted is somebody else's count of somebody else's list. Nobody can tell you exactly how many connectors exist. What everyone agrees on is that there are far more than you need, and no obvious way to tell which ones matter for you.

So users do one of two things. They connect nothing, and Claude stays a very good writing assistant that knows nothing about their business. Or they connect everything, and Claude becomes slower, vaguer, and weirdly bad at choosing the right tool. Both feel like the platform's fault. Neither is.

Here's the reframe that fixes it. Stop asking which connectors are good. Start asking what shape a working setup has. Because every genuinely useful Claude workflow, in any business, has the same shape, and once you can see it you can build your own stack in an afternoon.

Section 02

Every working setup is the same four links

Watch someone use Claude well and you'll see four moves, always in the same order. Raw material comes in from somewhere. Something applies judgment to it. Something builds the artifact. The artifact lands where it has to land to count as done.

01

Source Where the raw material already lives. Your inbox, your calendar, your store, your CRM, your file drive. Claude can't reason about your business from nothing.

02

Judgment Claude plus a skill pack that knows your trade. This is the link almost nobody fills on purpose, and it's the one that decides whether the output is any good. It is not a connector.

03

Make Where the artifact gets built. A deck, a graphic, a document, a page. Some jobs skip this link entirely because the artifact is just text.

04

Land Where the output has to end up before the work counts as finished. A draft in your inbox, a queued post, a page in your course platform, a row in your tracker.

Now the diagnosis. Go look at your own connector list. Most users have four or five things plugged into the Source link, nothing at all in Judgment, and nothing in Land.

That's why their AI work dies in a chat window. They can ask Claude anything about their business and get a smart answer, and then they copy the answer out by hand, paste it somewhere, and reformat it. The chain has a beginning and no end.

🎯

The fastest audit you can run: open your connector settings and sort what you've connected into the four links. If Land is empty, that's your next connector. Not another Source.

Section 03

The Judgment link is the one nobody connects

A connector moves data. That's its whole job, and it's an honest job. The Shopify connector can hand Claude your entire product catalog in a second. It has no opinion whatsoever about whether your product descriptions are any good.

That gap is where most disappointment with AI tools actually comes from. Users connect their data, ask for something, and get a competent, generic answer. The data was there. The judgment wasn't.

Judgment is what a skill pack supplies. A skill is a folder of instructions that gives Claude a specific way of working: what to ask before it writes, what a good version of this document looks like, what the common mistakes are, what to refuse to do. A plugin bundles a set of related skills together, along with the connectors they expect and any background agents that come with them.

The difference in output is not subtle. Ask Claude cold for a client proposal and you get a competent template. Ask through a delivery-focused skill pack and it asks about your rates first, warns you which parts of the scope are likely to cost you money, writes the proposal, writes the covering email, writes the follow-up sequence, and tells you honestly if a proposal is even the right next move.

Where skills and plugins actually run

This matters more than it looks, because the answer differs by surface and by plan. Here's the current position.

What Plans Where it runs

Skills Free, Pro, Max, Team, Enterprise Web, desktop, Claude Code, Cowork, and the Microsoft 365 add-ins

Plugins All paid plans Web chat, desktop chat, and Cowork

Directory connectors All plans Web, desktop, mobile, Claude Code, Cowork

Custom connectors All plans, but Free is capped at one Same, added by pasting an MCP server URL

Sub-agents and hooks Paid plans Cowork only. They appear greyed out in chat.

The line that catches people out is the last one. A plugin's skills work everywhere, but the background agents bundled with it only run in Cowork. If a pack advertises a weekly sweep that flags your overdue invoices, you get that sweep on demand in chat and automatically in Cowork. Same intelligence, different trigger.

🔑

Worth knowing: Cowork is now available on every paid plan, not just Team and Enterprise. On desktop it's on all paid plans, and on web and mobile it covers Pro, Max, and Team. If you dismissed it as an enterprise feature last year, that's out of date.

Section 04

Why more connectors makes Claude worse

This is the part nobody selling you an AI course wants to say. Every connector you enable costs something, and the cost is paid out of the same budget that holds your actual conversation.

When a connector is active, the descriptions of everything it can do get loaded so Claude knows what's available. A single connector might expose twenty or thirty separate actions, each with its own description. Connect eight of them and a meaningful slice of the working memory is spent before you've typed a word.

One engineering team measured this in a development setup and found that three fairly standard integrations consumed over 20 percent of the available context before the assistant started working. That was a code-heavy setup, which is heavier than most business chat, so treat it as an illustration rather than a law. The direction is what matters.

The symptoms are the frustrating kind, because nothing visibly breaks. Instructions get ignored. Answers drift generic. Claude reaches for the wrong tool, or asks you something it should already know. It feels like the model got worse. It didn't. It got crowded.

⚠️

Anthropic ships a setting for exactly this. Once you pass roughly ten active connectors, you can switch how they load from Auto to On demand, which stops everything loading at the start of every conversation. That setting existing at all is a quiet admission that connector count has a cost.

The four-connector rule

One connector per link. Four total. When you want a fifth, it replaces one rather than joining the set.

This sounds arbitrary and isn't. It's just the observation that a workflow with two Sources usually means you haven't decided where your real data lives, and a workflow with two Lands usually means you haven't decided what finished looks like. The constraint forces the decision.

You'll break it eventually, and that's fine. A seller running two storefronts genuinely needs two Sources. But break it deliberately, after the four-link version works, not by accident on your first afternoon of enthusiasm.

Section 05

Read or write decides what a connector can be

Here's the single most useful thing to know before you connect anything, and almost nobody says it out loud: most connectors are read-only.

Ahrefs exposes 61 tools through Claude and every one of them reads. Semrush has 14, all read. Helium 10 has 34, all read, and its own documentation is explicit that it cannot place a bid, pause a campaign, or edit a listing. Triple Whale states plainly that it cannot mutate your business data, ad accounts, campaigns, or budgets. None of that is a flaw. Reading is what those tools are for.

But it settles where each one can sit in your chain. A Source can be read-only. A Land cannot. If the only connectors you have are read-only, you don't have a chain, you have a very well-informed conversation that still ends with you copying text into another window.

🔍

Check this before you connect anything. Open the connector's page in the directory and read its tool names. If they all start with get, list, search, or analyze, it's a Source. If you see create, send, update, or publish, it can be a Land.

They don't fire on their own

A connector is not an automation. It does nothing until you start a conversation and ask. Buffer's official server, released in May 2026, is a good example: Claude can list your channels, see your queue, draft posts, and schedule them. What it cannot do is watch anything or react to an event. Buffer's own documentation is clear that the server only works inside a conversation you start. Gorgias says the same about its helpdesk connector: it will answer a ticket when you ask, and it will never notice one arriving.

If you want something to happen at 7am on Monday whether or not you're at your desk, a connector alone won't do it. That's what Cowork's scheduled tasks and background agents are for, and it's the strongest practical reason to care which surface you're working on.

Most writes land as drafts, but check which

Kajabi's server, included on every Kajabi plan at no extra cost, can build pages, write emails, and set up offers. Kajabi's marketing page says every write action lands as a draft. Kajabi's own technical documentation says otherwise, and the documentation is right: creating a page, course, or email does produce a draft, but community posts, direct messages, pinned announcements, navigation links, and theme edits all take effect immediately. On theme content the docs state it flatly, that changes take effect on published pages and there is no draft layer.

Mailchimp goes the other way and will draft a campaign but refuses to send it, so a human has to press the button. Typeform's connector will publish a form live and its documentation warns you to check it first. There's no universal rule here. Read the two paragraphs of documentation before you let a connector write anything.

Section 06

The connectors that aren't what their name says

This section will save you more time than any other. A connector's name tells you which company built it. It does not tell you what it does, and in several cases the gap is wide enough to waste an afternoon.

Twilio cannot send a text message The name says SMS. The connector has two tools, and both of them search Twilio's own API documentation. It never touches your account. If you want Claude to send a real text from your real business number, the connector that actually does it is Quo, formerly OpenPhone, which exposes send-message alongside call transcripts and contacts.

Yelp and Thumbtack are built for the customer, not the business Both are in the directory and both look like obvious wins for a local operator. They're search and discovery tools for people hiring a plumber. Neither lets a plumber manage a listing, see their leads, or reply to anything.

Birdeye cannot reply to a review Birdeye sells reputation management. Its connector has around 30 tools and all of them read: reviews, listings, sentiment, competitors. Useful for understanding where you stand, useless for doing anything about it. Also worth knowing that birdeye.com and birdeye.so are different companies, and the second one is a crypto data platform that also ships a connector.

Yotpo's connector is not the Yotpo you're thinking of Sellers know Yotpo for reviews, loyalty, and SMS. The connector is for Yotpo Discover, a separate product that tracks whether AI answer engines cite your brand. It cannot respond to a review or touch your loyalty program.

HyperFrames by HeyGen makes no avatars HeyGen is known worldwide for talking-head AI video. This connector renders motion graphics and animated slides from HTML and CSS. It's genuinely good at that. It is not what anyone searching for HeyGen expects to find.

There are two QuickBooks connectors and they are not equals Intuit's own is mostly reporting: profit and loss, cash flow, benchmarking. A separate free third-party one called Meridian exposes 63 tools with full create and update access across invoices, bills, customers, vendors, and expenses. Same brand on the box, very different machine inside.

A press release is not the directory In August 2026, HoneyBook, Pipedrive, and Bonsai all published announcements describing live Claude connectors. Searching the directory by name returns nothing for any of the three. The announcements may be early, or the rollout may be partial. Either way, the lesson generalizes: check Settings, then Connectors, and search the name yourself before you plan a workflow around a headline.

🎯

The flip side is better news. The directory goes far deeper into specific trades than most people expect. There is a connector for a roofing CRM. Search your own trade by name before you assume the tools you use every day aren't there.

Section 07

Five stacks you can copy today

Each stack fills all four links with tools built for that trade, names the pack that supplies the judgment, and lists the specific skills the chain runs through. Find your row, then read that one card.

Find your stack

You post content for a living SparkToro→30 Days of Content→Canva→Metricool

You run a local service business Local Falcon→The Local Business Magnet→Canva→Quo

You deliver client work solo Fathom→The Client Delivery Operator→PandaDoc→Harvest

You sell physical products online Shopify→The E-Commerce Seller→Canva→Klaviyo

You build and sell courses Riverside→Course Creator→Gamma→Kajabi

Connector a tool you connect skill or pack the judgment layer Reads Source only Writes can be a Land

Stack 01 The content creator

The chain

SourceSparkToroPaste URL

→

Judgment30 Days of ContentPlugin

→

MakeCanva1-click

→

LandMetricoolPaste URL

The connectors, precisely

SparkToro ReadsAudience research. Tells you what your audience actually reads, watches, and follows, across sites, podcasts, YouTube, and subreddits. The only tool here built for the question a creator asks every week. Custom connector.

Canva Writes33 tools. Generates and edits real designs in your account, then exports. It does not post them anywhere.

Metricool WritesSchedules to 11 organic networks, reads analytics back, tracks competitors. Works on the free tier, capped at 30 days of data and 20 scheduled posts a month. Custom connector.

Worth adding laterAhrefs for its brand-radar tools, which track whether AI answer engines are citing you. That is the most 2026 thing in the entire directory. OpusClip if you make long video: it cuts clips, burns captions, and reframes to 9:16. Paid plan only.

The weekly job, step by step

SparkToro tells you what your audience is actually consuming this month

content-pillar-builder turns that into pillars grounded in real audience behavior, not guesswork

content-month-architect maps thirty days against those pillars

hook-lab writes ten opening variants for the five posts that matter most

visual-prompt-director briefs the graphics, and Canva builds them

Metricool schedules week one and reports back how last month's posts did

The skill pack

30 Days of Content 15 skills. This chain runs through four of them.

content-pillar-builder content-month-architect hook-lab visual-prompt-director

Runs on: all surfaces. Needs Cowork for: running the monthly batch on a schedule rather than remembering to ask.

The gap: there is no official Reddit, YouTube, or TikTok connector. Three of the most obvious places a creator would want to pull from, and none of them are there. Metricool covers the posting side, but research on those platforms is still manual.

Stack 02 The local service business

The chain

SourceLocal Falcon1-click

→

JudgmentLocal Business MagnetPlugin

→

MakeCanva1-click

→

LandQuo1-click

The connectors, precisely

Local Falcon Writes37 tools. Runs geo-grid rank scans that show where you appear on the map from different points in your service area, manages tracking campaigns, and watches your business profile for unauthorized changes. The single most on-the-nose connector for a local operator. Needs the Basic plan, around $50 a month.

Quo WritesFormerly OpenPhone. Sends real texts from your real business number, reads call transcripts, manages contacts. This is the connector that closes the loop a local business actually runs on. From $19 per user a month, no free tier.

Canva WritesProfile photos sized correctly, referral cards, seasonal promo graphics.

Worth adding laterMeridian for QuickBooks, which is free and has 63 tools including invoice and expense creation. Or Square if that's your payment rail, though its connector is a raw API pass-through rather than a curated tool set.

The weekly job, step by step

Local Falcon runs a scan and shows where you rank on the map grid across your service area

google-profile-optimizer reads the result and tells you which profile fields are costing you position

review-request-engine writes one request per customer who finished this week, pitched to how that job went

Quo sends them as texts from your business number, which is where they actually get read

review-reply-writer handles the week's new reviews, and you paste those into your profile yourself

The skill pack

The Local Business Magnet 12 skills. This chain runs through three, and a fourth handles quoting.

google-profile-optimizer review-request-engine review-reply-writer quote-and-estimate-builder

Runs on: all surfaces. Needs Cowork for: the Friday sweep running itself.

The gap: there is still no official Google Business Profile connector, so nothing posts your review replies or updates your profile. Third-party bridges exist that hold Google's approved API access and will publish on your behalf, and they work, but you are handing a smaller company your profile authorization. Keep a human reading everything before it posts. Google can suspend a profile over automated content.

Stack 03 The consultant or freelancer

The chain

SourceFathom1-click

→

JudgmentClient Delivery OperatorPlugin

→

MakePandaDoc1-click

→

LandHarvestPaste URL

The connectors, precisely

Fathom ReadsMeeting intelligence. Hands Claude the full transcript of a client call, not just a summary, plus search across every meeting you've had. A consultant's raw material is what was said on calls, which makes this the most natural Source in this whole article.

PandaDoc WritesBuilds a document straight from markdown and sends it for signature. Free tier allows five signatures a month; unlimited sending from about $19 a month.

Harvest WritesStarts and stops timers, logs time, creates clients and projects, drafts invoices. The connector that turns work into money owed. Custom connector.

Pick one notetaker, not threeGrain, Fireflies, tldv, Fellow, and Granola all have connectors and all hand over transcripts. Granola gates transcripts behind a paid plan, so its free tier gives you summaries only. The second notetaker adds nothing but tool clutter.

The weekly job, step by step

Fathom supplies the full transcript of every client call you had this week

weekly-client-update writes the right email for the kind of week it was, quoting what was actually agreed on the call

scope-creep-defender catches the request buried at minute 34 that nobody wrote down and nobody priced

PandaDoc turns the resulting change order into a document and sends it for signature

Harvest logs the hours discussed and drafts the invoice

The skill pack

The Client Delivery Operator 12 skills plus a background agent that sweeps your client tracker.

weekly-client-update scope-creep-defender change-order-writer late-payment-chaser client-radar (agent)

Runs on: the skills work everywhere. Needs Cowork for: the client radar agent, which sweeps for silent clients and overdue invoices without being asked.

The gap: HoneyBook, Pipedrive, and Bonsai all announced connectors in August that don't appear in the directory when you search by name. If you run your business on one of those, check yourself before you plan around it. Toggl and Clockify have nothing yet either, and Clockify's own page says its server is still coming.

Stack 04 The e-commerce seller

The chain

SourceShopify1-click

→

JudgmentThe E-Commerce SellerPlugin

→

MakeCanva1-click

→

LandKlaviyo1-click

The connectors, precisely

Shopify Writes25 tools with real create and update access to products, orders, customers, discounts, and inventory. The spine of this stack and the deepest official commerce connector that exists.

Klaviyo Writes27 tools including campaign creation and template assignment, plus flow performance and metrics. Where most solo-seller revenue per hour comes from.

Triple Whale ReadsThe sharpest read on whether your ad spend is working. Read-only by design, which is a feature. Needs the Foundation plan or above. Custom connector. Windsor.ai is the honest budget alternative with a real free tier.

Worth adding laterGorgias for support replies and macros, currently in beta. Shippo to compare live carrier rates and buy labels. Helium 10 if you sell on Amazon, though it's read-only and needs their Diamond tier at around $279 a month.

The morning job, step by step

Shopify gives yesterday's revenue, orders, and stock levels

Triple Whale says whether the spend behind those orders actually paid for itself

daily-store-digest tells you the two or three things that need attention, not the numbers you can already see

inventory-forecaster flags the SKUs about to stock out, and supplier-negotiation-assistant drafts the reorder email

email-flow-architect fixes the flow that's underperforming, straight into Klaviyo

The skill pack

The E-Commerce Seller 15 skills. The morning chain uses four of them.

daily-store-digest inventory-forecaster supplier-negotiation-assistant email-flow-architect listing-seo-optimizer ad-creative-generator

Runs on: all surfaces. Needs Cowork for: the digest arriving daily instead of when you think to ask.

The gap, and it's the biggest one in this article: Etsy has no connector, official or vendor-built, and Etsy's current API removed shop analytics entirely, so nobody could build a good one today. Amazon Seller Central has nothing either. Amazon Ads has an official server but it's in beta and needs partner API credentials. If you sell anywhere other than Shopify, the connector ecosystem is not yet built for you, and every skill in the pack is designed to work from a pasted export instead.

Stack 05 The course creator

The chain

SourceRiverside1-click

→

JudgmentCourse CreatorPlugin

→

MakeGamma1-click

→

LandKajabiPaste URL

The connectors, precisely

Riverside WritesOver 50 tools. Searches your recording archive by transcript, creates new edit versions without touching originals, cuts clips, and schedules to social. Needs the Grow plan or above, and the feature is still in beta. Descript is the alternative if you're refining footage you already have rather than recording fresh.

Gamma Writes18 tools. Turns a lesson outline into a real deck, plus image generation and viewer analytics. Works on the free plan, no API key needed.

Kajabi WritesCourses, offers, landing pages, emails, blog posts. Free on every Kajabi plan. Creates land as drafts, but community posts, DMs, announcements, navbar and theme edits go live immediately. Custom connector.

If you're not on KajabiThinkific has an official server but it is read-only and needs their Plus plan. Teachable's is experimental, waitlisted, and runs locally. Skool has nothing official, and the community projects work by extracting your session cookie, which risks your account. Don't.

The module build, step by step

Riverside searches your recording archive and finds the twelve minutes where you already explained this concept well

curriculum-architect has already fixed what this module must make the student able to do

lesson-scripter writes to that outcome rather than to a word count

slide-deck-builder produces the structure, and Gamma generates the actual deck

workbook-and-worksheet-maker produces the companion worksheet

course-sales-page-writer drafts the page, and Kajabi builds it as a draft you review before it goes live

The skill pack

Course Creator 14 skills. A module build runs through four, and launch adds a fifth.

curriculum-architect lesson-scripter slide-deck-builder workbook-and-worksheet-maker course-sales-page-writer

Runs on: all surfaces. Needs Cowork for: nothing essential. This is the stack that works most completely in ordinary chat.

The gap: step 1 depends on you having recorded in Riverside. If your archive is a folder of files, Descript can import them but cannot export finished video through the connector, so you'll finish in the app. Nobody has solved the "folder full of old footage" problem well yet.

Section 08

Build your own stack in twenty minutes

None of those five is your business exactly. That's fine, because the method is the transferable part, and it runs backwards from the way most people approach it.

Don't start with the directory. Start with the deliverable you resent making.

The twenty-minute build 1. Name the job. Pick one thing you make every week that you'd pay someone else to do. Not a category. The actual artifact: the Friday client update, the Monday store check, the eight captions.

2. Ask what it's made of. What do you have to go and look at before you can write it? That's your Source, and there's usually only one that matters. Connect that.

3. Ask where it has to end up. Not where you'd like it, where it has to be before the job is done. Inbox, queue, tracker, platform. That's your Land. Check its tool list writes before you count on it.

4. Ask whether it needs building. If the artifact is a deck or a graphic, you need a Make link. If it's text, you don't. Plenty of good stacks are three links.

5. Fill Judgment last, and on purpose. Find the pack that covers this trade. If none exists, write the standard down yourself: what you'd tell a new hire about what good looks like here, in a page, saved where Claude can read it.

6. Run the job once, end to end. Not a test. The real one, this week. Whatever breaks is your next connector.

Step six is the one people skip, and skipping it is why so many carefully built setups get abandoned. You cannot tell whether a stack works by looking at it. You find out on the Friday when you're tired and you'd rather not.

⚡

The honest test for any connector: does it remove a copy and paste you're currently doing by hand? If yes, connect it. If you can only describe the benefit in general terms, leave it in the directory.

Section 09

What to do this week

Open your connector settings and count what's there. Sort each one into Source, Judgment, Make, or Land. Most users find three or four Sources, an empty Land, and a couple of things they connected months ago out of curiosity and have never used once.

Turn off everything you can't attach to a job you actually do. This will feel like losing capability. It isn't. Every connector you remove gives Claude back room to think about the ones you kept.

Then fill the empty link. If it's Land, that's a fifteen-minute fix and the largest single improvement available to you, because it's the difference between advice and finished work. Check the tool list writes before you commit. If the empty link is Judgment, install the pack that covers your trade and run one job through it.

The users getting real work out of Claude right now are not the ones with the most connectors. In most cases they have noticeably fewer than the people struggling. What they have is four that form a chain, and a clear idea of which weekly job that chain exists to kill.

Pick your job. Build the four links. Run it once this Friday.