DEV Community: Built2WinWeb

Keeping a Website Healthy: What Maintenance Covers, Who Should Do It

Built2WinWeb — Sat, 24 Oct 2026 07:38:12 +0000

A website isn't finished the day it launches. The software underneath keeps changing, certificates and domains expire on a schedule, hosting occasionally fails and forms break quietly when something upstream changes. This guide covers what maintenance actually includes, what skipping it costs, a schedule you can keep yourself and how to judge a plan if you'd rather hand it off.

I define maintenance as everything that keeps an existing site safe, working and fast, as distinct from redesigns and from hosting. Almost every plan or sensible routine comes down to six jobs: backups (stored off-site and actually test-restored), updates, security, monitoring, renewals and fixes. How much a site needs depends mostly on how many moving parts it has.

The costs of skipping it are ordinary rather than dramatic: a known vulnerability gets exploited, a contact form stops delivering and nobody notices, a certificate or domain lapses, speed slowly decays, a forced PHP upgrade lands all at once, or there's no clean backup when you need one. The schedule covers continuous, monthly, quarterly and yearly tasks.

I list the signs it's time to hand maintenance off and eight questions to ask before signing up for any plan, including whether you keep your logins and backups if you cancel. There's a disclosed look at BugShield's WordPress plans as a content partner, and a note on why custom-coded sites with no plugin stack need less upkeep.

Key takeaways

Maintenance is six jobs: backups, updates, security, monitoring, renewals and fixes
A backup you've never restored isn't a plan
Test every form monthly by submitting it
Track domain and certificate renewals and the card behind them
Before signing a plan, confirm you keep your logins and backups

Related on BuiltToWinWeb: BuiltToWinWeb's services

Read the full guide on BuiltToWinWeb: The complete website maintenance guide

A summary by Jacob Campbell of an article first published on built2winweb.com on October 8, 2026.

I Made My Site Fully Agent-Readable. Here Are the Real Numbers.

Built2WinWeb — Sat, 01 Aug 2026 00:00:00 +0000

I Made My Site Fully Agent-Readable. Here Are the Real Numbers.

🤖 Agent readiness · Updated August 2026 · 9 min read · by Jacob Campbell

Over the past few months I turned built2winweb.com into about as agent-readable a site as a solo developer can build: an MCP server, an A2A agent card, an NLWeb /ask endpoint, ACP, UCP, agents.txt, llms.txt, an RSL licence, Markdown content negotiation on every page, and IndexNow wired to the CDN. Then I sat down and actually measured it. This is the honest version — including the parts that did not work, the two Cloudflare defaults that were fighting me, and the number that matters more than all the protocol plumbing combined.

By the numbers

209,600 — URLs pushed to IndexNow
1,319 — Pages actually indexed by Bing
28,685 — URLs in the sitemap
16 — Referring domains — the real ceiling

What I actually built

The full surface is documented on the agent-ready page, but briefly: an MCP server at /mcp exposing seven tools, a SEP-2127 server card, an A2A agent card with the AP2 merchant extension, an ACP descriptor, a UCP commerce profile, per-path agent permissions, an /ask endpoint running BM25 over the site with Schema.org-typed responses, llms.txt and llms-full.txt, AGENTS.md, a Markdown sitemap, and an RSL 1.0 licence referenced from robots.txt that explicitly permits search indexing, RAG and model training.

Every one of those is live, public and needs no API key. None of it uses a framework — it is plain PHP on shared hosting. If you want to see it, call it.

What happened: the numbers

Here is where honesty is more useful than a case study. Across roughly three months:

209,600 URLs submitted to IndexNow
1,319 pages indexed in Bing
28,685 URLs in the sitemap
111 impressions, 0 clicks in Bing search
16 referring domains

Seznam tells the same story from another angle: 3,390 URLs discovered, 86 kept in its crawler database, 19 indexed. That is 0.6% of what it found.

So: I notified search engines about my pages roughly seven times over, and about 4.5% of the site is indexed. Notification was never the constraint.

Cloudflare Crawler Hints was firing 4,900 URLs every 18 hours

This is the part I would not have caught without opening Bing Webmaster Tools. Every single one of those 209,600 IndexNow submissions came from Cloudflare Crawler Hints , which I had enabled and forgotten. On a 28,685-URL site it was pushing 4,900 URLs every 18 hours — including pages I had deliberately marked noindex, and at least one .well-known challenge file that returns 404.

Bing noticed. Its top recommendation on my dashboard read: "Avoid IndexNow Batch Mode to prevent excessive server load and potential indexing delays." I had managed to turn a helpful protocol into a self-inflicted problem by leaving a toggle on. IndexNow is for content that changed. Firing your whole site at it repeatedly is how a key gets deprioritised.

My CDN was contradicting my own robots.txt

The second surprise was worse, because it worked directly against the thing I was trying to do. My robots.txt spends about 200 lines explicitly welcoming AI crawlers — GPTBot, ClaudeBot, PerplexityBot, CCBot and the rest — with Content-Signal: search=yes, ai-input=yes, ai-train=yes.

Above all of that, Cloudflare was injecting a managed block that said the opposite:

# BEGIN Cloudflare Managed content
User-agent: ClaudeBot
Disallow: /
User-agent: GPTBot
Disallow: /
User-agent: CCBot
Disallow: /
User-agent: Google-Extended
Disallow: /
# END Cloudflare Managed Content

It also declared ai-train=no where mine said yes. Under RFC 9309 duplicate user-agent groups get merged and Allow: / wins a tie, so well-behaved crawlers probably still fetched — but I was relying on tie-breaking to undo a directive I never wrote. Check your rendered robots.txt, not your source file. Your CDN may be editing it.

Discovery and citation are different problems

Here is the conclusion I did not want but the data insists on. Every protocol I implemented solves discovery — helping a machine find and parse what exists. None of them solve authority — giving anyone a reason to cite it.

With 16 referring domains, Bing is choosing not to index most of my site, and no amount of notification changes that decision. Bing said so directly in its second recommendation: "Your site does not have enough inbound links from high quality domains." Two search engines, zero crawl errors, zero guidelines violations, and still almost nothing indexed. That is not a technical failure. It is a judgment about whether the content is worth storing.

If you are considering this work as an SEO play, do not. It is not one.

The bug that was hiding underneath all of it

While auditing, I found something more embarrassing and more consequential than any protocol detail. Thirteen of my article URLs were serving the wrong article.

My rewrite rules pointed several slugs at positional anchors — blog-card-1, blog-card-3 — that referred to slots on the blog index rather than to specific posts. When the index got reordered, the slugs silently detached from their content. A URL about migrating from Shopify was serving an article about critical CSS. One of them was in my sitemap.

Every anchor I had named descriptively was still correct. Not one had drifted. The lesson is boring and universal: name things after what they are, not where they sit. And an index full of pages that do not deliver what their title promises is a very good explanation for a 0% click-through rate.

What I would tell you to do

If you want to make a site agent-readable, the order that actually matters:

Fix what is broken first. I found mismatched routes, 2,020 stale 404s from apostrophes in city slugs, and a CSS rule that had been rendering a page white-on-white. Those cost more than any missing manifest.
Publish llms.txt and an /ask-style endpoint. Cheap, genuinely useful, and lets an assistant get a cited answer instead of guessing.
Check what your CDN is doing on your behalf. Both of my biggest problems were defaults I never chose.
Then earn links. This is the slow one, and it is the one that decides everything else.

So was it worth doing?

Yes — but not for the reason I started. It bought precision, not traffic. When an assistant is asked what a custom site costs, it can now fetch a typed answer with a citation instead of paraphrasing marketing copy. Nothing is generated on my origin, so nothing about my pricing gets hallucinated. That is worth something on its own, and it is a reasonable bet on how people will find things next.

What it is not is a shortcut. There is no manifest you can publish that substitutes for being worth citing. I would rather tell you that plainly than sell you the version where the graph goes up.

If you want a site built this way

I hand-code every site I build in PHP — no WordPress, no page builders, no monthly fees — and the agent surface above is the same kind of work. If you want a fast site that both people and machines can actually read, get a free quote and tell me what you need.

Frequently asked questions

Did adding MCP, A2A and llms.txt improve search rankings?

No. Across three months the site went from a handful of indexed pages to 1,319 out of 28,685, with 111 impressions and 0 clicks in Bing. These protocols solve discovery — helping machines find and parse content. They do not create authority, and authority is what decides whether a search engine bothers to index you.

What is Cloudflare Crawler Hints and should I turn it off?

It automatically submits changed URLs to IndexNow on your behalf. On my site it pushed 209,600 URLs, about 4,900 every 18 hours, including noindex pages and a 404. Bing flagged it as excessive. If your site is large or template-generated, check what it is actually submitting before leaving it on.

Why would my CDN change my robots.txt?

Cloudflare can inject a managed block that disallows AI crawlers, and it appears above your own rules. Mine was blocking GPTBot, ClaudeBot, CCBot and Google-Extended while my own file explicitly welcomed them, and declared ai-train=no against my ai-train=yes. Always read the rendered file at your live URL, not the source on disk.

Is IndexNow worth setting up?

Yes, for genuinely changed content — it is a key file plus one HTTP POST, and Bing, Yandex, Naver and Seznam all consume it. Google does not. Just do not bulk-resubmit an unchanged site; that is the behaviour that gets a key deprioritised.

What should a small site do first?

Fix what is broken, publish llms.txt and a question-answering endpoint, audit what your CDN is doing by default, and then spend your remaining effort earning links. The technical work is a weekend. The links are the years.

Want a site machines can actually read?

I hand-code fast PHP sites with the structured data, schema and agent endpoints built in — one flat fee, no subscriptions, you own the code.

Get my free quote

Turn Your Website Into a Lead-Generating Machine

Built2WinWeb — Fri, 10 Jul 2026 00:00:00 +0000

How to Turn Your Website Into a Lead-Generating Machine

🚀 More Leads · Updated July 2026 · 9 min read · by Jacob Campbell

Most business websites are digital business cards: they sit there, look okay, and do nothing. A lead-generating website is different — it works. It shows up when people search, loads instantly, earns trust in seconds, and makes the next step impossible to miss. The gap between the two is not luck or budget; it is a handful of deliberate choices. Here is exactly what turns a passive site into one that fills your inbox with real customers.

Key facts

Found — A lead machine shows up in search, not just when you share it
Fast — Under-1s loads keep the visitors you worked to get
One CTA — A single, obvious next step on every page
Proof — Reviews and real work turn browsers into buyers

It gets found in the first place

A site cannot generate leads from people who never reach it. The first job is visibility: showing up in Google when people in your area search for what you offer. That means a site built for SEO from the ground up — fast, properly structured, marked up with the right data, and targeting the terms real buyers use. This traffic is the best kind: free, ongoing, and made of people already looking to buy. Everything else on this list only matters once customers can actually find you.

It loads before they lose patience

You fought to earn a visitor — do not lose them to a spinner. Speed is where most sites hemorrhage leads, because people abandon slow pages in seconds and never come back. A lead-generating site loads in under a second, so the momentum that brought someone in carries them to the contact form instead of the back button. Speed is not a technical vanity metric; it is the difference between a visitor who converts and one who never sees your offer.

It makes the next step obvious

Confused visitors do not convert. Every page needs one clear action — call, book, request a quote, buy — stated plainly and repeated. Not five competing links, not a phone number buried in the footer, not a form with fifteen fields. The best lead machines are almost aggressively clear: here is what we do, here is the proof, here is the button. When the path to becoming a customer is effortless, far more people take it. Clarity converts; clever converts nobody.

It earns trust fast

People buy from businesses that feel safe. Reviews, real photos of your work, clear pricing, a visible human, recognizable logos — these are the signals that flip a nervous visitor into a confident buyer within seconds. A generic template with stock photos and no proof does the opposite. You do not need to overwhelm people; you need to answer the quiet question every visitor is asking — 'can I trust these people with my money?' — before they have to ask it.

It captures the lead cleanly

When someone is ready, make it frictionless. A short, mobile-friendly contact form, a tap-to-call button on phones, an instant quote request — every extra field or hoop loses people at the finish line. And a lead is only useful if it reaches you: forms should land in your inbox reliably, not vanish into a plugin. The goal is a straight, short line from 'interested' to 'in your inbox,' with nothing clever in the way to trip a ready buyer.

It’s built to keep working

A lead machine is not a one-time setup; it is a site that keeps performing without constant babysitting. That is far easier when it is built on clean, fast code you own rather than a fragile stack of plugins on a monthly plan. Add a bit of fresh content over time — like the guides you are reading now — and the search traffic compounds. The best websites quietly get better at generating leads the longer they run, instead of decaying until the next expensive redo.

Want a website that actually sells?

Getting found, loading instantly, one clear action, real proof, and a clean path to your inbox — that is the recipe, and it is exactly what I build. Not a digital business card, but a fast, hand-coded, SEO-ready site designed to turn visitors into customers, owned by you for one flat fee. If your current site is just sitting there, it is time it started working. Get a free quote and let’s build a lead machine.

Frequently asked questions

Why isn’t my website generating any leads?

Usually one of five reasons: people can’t find it in search, it loads too slowly, there’s no clear next step, it doesn’t build trust, or the contact path is clumsy. Fix those and a passive site starts producing real inquiries.

What’s the single most important factor?

Getting found, then keeping them. A site has to show up in search and load in under a second — otherwise the rest never gets seen. After that, one clear call to action does the heavy lifting.

How many calls to action should a page have?

One primary action, repeated. Multiple competing links dilute focus and stall visitors. The clearest lead-generating pages make a single next step impossible to miss.

Do I need reviews to generate leads?

Trust signals like reviews, real photos of your work, and clear pricing dramatically increase conversions because they answer the visitor’s quiet question — can I trust these people? — before they leave to look elsewhere.

Can my current website be turned into a lead machine?

Sometimes with targeted fixes, but sites on slow, plugin-heavy platforms often need a rebuild to get the speed and SEO that lead generation depends on. A quick review will tell you which path makes sense.

Turn your website into a lead machine

I build fast, hand-coded, SEO-ready websites designed to get found and turn visitors into customers — one flat fee, fully owned by you.

Get my free quote

OBS Studio MCP: Run Flawless Class Streams Without the Panic

Built2WinWeb — Fri, 10 Jul 2026 00:00:00 +0000

OBS Studio MCP: Run Flawless Class Streams Without the Panic

🎥 OBS & MCP · Updated July 2026 · 8 min read · by Jacob Campbell

Recording and streaming lessons is now part of the job, and OBS Studio is the free tool everyone recommends — right up until you open it and face a wall of scenes, sources, and audio routing. OBS MCP takes the fear out of it. It connects Claude to OBS through the OBS WebSocket protocol, so you can switch scenes, add sources, and start recording by simply asking — and, more importantly, learn how OBS is organized by watching it done. For a teacher, that means clean recordings and calm live streams without a semester of trial and error.

Key facts

WebSocket — Claude controls OBS through the built-in OBS WebSocket
OBS 31+ — Needs OBS Studio 31 or newer with WebSocket enabled
Hands-free — Switch scenes and start recording by asking
Free — OBS and the MCP server are free and open source

What OBS MCP does

OBS MCP is an open-source MCP server (royshil/obs-mcp) that controls OBS Studio through the OBS WebSocket protocol. Once you enable the WebSocket server inside OBS and connect the MCP server to Claude, Claude can create and switch scenes, add and toggle sources, manage audio, and start or stop recording and streaming. You are still running OBS on your machine — Claude is just operating the controls you would otherwise hunt for, and telling you what it is doing as it goes.

Why this fits teachers so well

Most educators do not want to become streamers — they want a reliable recording of today's lesson and a stream that does not fall apart mid-class. OBS's power is exactly what makes it intimidating. With OBS MCP you can say 'set up a scene with my webcam in the corner and my slides full screen, then start recording,' and watch it happen. You get a working setup immediately, and because Claude names each piece — scene, source, filter — you quietly learn how OBS actually thinks.

Learn the mental model, not the menus

OBS becomes simple once you understand three ideas: scenes, sources, and the audio mixer. The MCP is a fast way to internalize them. Ask Claude to build a 'lecture' scene and a 'screen-share' scene, then switch between them, and you see how scenes are just saved arrangements of sources. Ask why your mic is picking up echo and have it walk through the audio settings. In one sitting you move from 'this is overwhelming' to 'I know where everything lives' — the exact confidence you need before recording anything real.

Set up a class recording in minutes

Try this before your next lesson: enable OBS WebSocket, connect OBS MCP to Claude, and ask it to create a scene with your camera, your slides, and clean audio, then start recording. Deliver the lesson, then ask it to stop and tell you where the file saved. Next time, ask for a second scene that shows a document camera or a demo app, and switch to it mid-lesson with a sentence. You have built a repeatable recording kit — and learned enough OBS to adjust it yourself.

Staying in control during a live class

For live streams, rehearse once with the MCP so nothing is a surprise: have it pre-build your scenes, confirm audio levels, and do a short test recording. During class you can keep driving by voice-of-text — 'switch to the screen-share scene' — instead of clicking around while thirty students watch. Keep a manual fallback in mind (know the scene hotkeys), connect only your own OBS, and treat streaming/recording toggles as deliberate actions you confirm. Calm comes from having practiced, and the MCP makes practice quick.

The payoff for your students

Better recordings and steadier streams are not vanity — they are access. A clean, well-framed lesson is one a student can actually rewatch, a remote learner can follow, and an absent kid can catch up on. When the software stops being the obstacle, you spend your energy on teaching, not troubleshooting. And once you understand OBS well enough to guide it, you can teach students to record and present their own work — a genuinely useful skill they will carry beyond your class.

Sources & further reading

Frequently asked questions

What is OBS MCP?

OBS MCP is an open-source MCP server that lets Claude control OBS Studio through the OBS WebSocket protocol — switching scenes, managing sources and audio, and starting or stopping recording and streaming.

What do I need to run it?

OBS Studio 31 or newer with the WebSocket server enabled, and the OBS MCP server connected to Claude (you set a WebSocket password so only you can control it).

Do I need to be technical?

No. You describe the setup you want — a scene with your webcam and slides, then start recording — and watch it happen. It doubles as a fast way to learn how OBS scenes, sources, and audio work.

How does it help teachers specifically?

It gives you reliable recordings and calmer live streams without a long learning curve, and it teaches you OBS's mental model by narrating each step — so you can adjust setups yourself and even teach students to record their own work.

Is it safe?

You enable WebSocket with a password and connect only your own OBS. Rehearse once before a live class, keep manual scene hotkeys as a fallback, and treat recording and streaming toggles as deliberate actions.

Want a home for your lessons and recordings?

I build fast, custom sites — course hubs, video libraries, class pages — with the AI workflows to keep them updated. One flat fee.

Get my free quote

Roblox Studio MCP: Learn Game Design Before You Teach It

Built2WinWeb — Fri, 10 Jul 2026 00:00:00 +0000

Roblox Studio MCP: Learn Game Design Before You Teach It

🎮 Roblox & MCP · Updated July 2026 · 9 min read · by Jacob Campbell

Roblox is where a huge share of your students already spend their time — and increasingly where they want to build, not just play. The catch for teachers is Roblox Studio: a real game engine with a scripting language (Luau), a data model, and a learning curve. Roblox's own open-source Studio MCP server closes that gap. It lets Claude work directly inside Studio — inserting objects, editing properties, and writing scripts — while you watch and learn. For an educator, it is the fastest honest path to leading a game-design class: master the engine yourself, by building in it, before day one.

Key facts

Official — An open-source MCP server released by Roblox
In Studio — Claude acts inside your actual Roblox Studio place
Luau — See scripts written and explained, not just pasted
Free — Roblox Studio and the MCP server cost nothing

What the Roblox Studio MCP is

In 2025 Roblox released an open-source Studio MCP server — an official bridge that lets large language models like Claude modify experiences directly inside Roblox Studio. It runs alongside Studio and exposes its capabilities through the Model Context Protocol, so Claude can insert models and parts, set properties, organize the data model, and generate Luau scripts on request. You connect it once; after that you build your game by describing it, and the changes appear in the Studio you already have open.

Why teachers are the perfect users

A game-design teacher does not need to become a shipping engineer — they need to understand the engine well enough to guide, unblock, and grade. That is exactly what the Studio MCP accelerates. Ask it to 'add a spawn point, a moving platform, and a script that gives points when a player touches the goal,' and watch it wire the parts and write the Luau. You end up understanding the data model and the scripting patterns because you saw them assembled — the same patterns your students will hit in week one.

Watch the Luau, learn the logic

Scripting is where most new teachers stall, and where the MCP earns its keep. When Claude writes a Luau script through Studio, ask it to explain each block: what a service is, why it used an event, how the touched signal fires. You are reading real, working code in the context of your own game — far stickier than a generic tutorial. Do this across a few small builds and you will recognize the handful of patterns that power most beginner Roblox projects, and be able to teach them plainly.

A starter project for your class

Build an obby (obstacle course) tonight. Ask Claude, through the Studio MCP, to lay out a start pad, three jumping platforms, a hazard that resets the player, and a finish that shows a win message. Watch each object appear and each script get written. Then hand your students the same brief and let them extend it. You now have a working example, a graded rubric you actually understand, and — because you built it live — the confidence to answer 'why isn't mine working?' in real time.

Keeping it classroom-safe

Work in a test place, not a published experience with live players, and save versions as you go. Review scripts before you run them — reading them is the point, not skipping them. Because Roblox built the server, it is designed to operate within Studio's own permissions, but the same discipline applies: connect only the place you intend, and treat generated code as a draft to understand rather than a black box to trust. That habit is also the exact lesson you want students to learn.

From confident teacher to confident class

The goal was never to have AI make the games for your students — it is to make you fluent enough to teach well. When you can open Studio, describe a mechanic, and narrate what appears, your class stops being a nervous read-along and becomes a live demo. Students learn more from a teacher who can build and explain on the spot than from one reciting slides. The Studio MCP is how you get there in evenings, not semesters.

Sources & further reading

Frequently asked questions

Is there an official Roblox MCP server?

Yes. Roblox released an open-source Studio MCP server that lets LLMs such as Claude modify experiences directly inside Roblox Studio — inserting objects, editing properties, and writing Luau scripts.

Do I need to know Luau or Studio first?

No. You describe what you want and watch Claude build it in Studio, explaining the scripts as it goes. That is a strong way to learn Luau and the Roblox data model, because you see correct patterns in the context of your own game.

What does it cost?

Roblox Studio is free, and the Studio MCP server is open source. You connect it to Claude, which you already use.

How does this help me teach?

It lets you master the engine fast, prepare working example projects, and demonstrate mechanics live in class — so you can answer real questions instead of reading from slides.

Is it safe for a classroom?

Work in a test place, save versions, and review generated scripts before running them. Treat the code as a draft to understand — which is also the discipline you want students to build.

Running a game-design or coding program?

I build fast, custom websites for classes, clubs, and studios — plus the AI workflows around them. One flat fee, no monthly platform tax.

Get my free quote

Chrome DevTools MCP: The Must-Have for Search & SEO Issues

Built2WinWeb — Fri, 10 Jul 2026 00:00:00 +0000

Chrome DevTools MCP: The Must-Have for Search & SEO Issues

🔎 Chrome & SEO · Updated July 2026 · 9 min read · by Jacob Campbell

Most SEO problems are invisible in a text editor. A page looks fine, yet it will not rank — and the real cause is buried in how Chrome actually renders it: a slow Largest Contentful Paint, a layout shift, a blocked script, a resource that never loads. Chrome DevTools MCP — Google’s official MCP server — puts that whole diagnostic layer in Claude’s hands. It drives a real Chrome, measures performance, reads the console and network, and reports back. If you care about search, it is the closest thing to giving your AI the same eyes a senior technical-SEO would use.

Key facts

Official — Built and maintained by the Chrome DevTools team
Real Chrome — Drives an actual browser via Puppeteer — not a guess
CWV — Measures Core Web Vitals the way Google does
Console — Reads errors, network, and the rendered DOM

What Chrome DevTools MCP is

Chrome DevTools MCP is Google’s official Model Context Protocol server that gives an AI assistant the full power of Chrome DevTools. Under the hood it uses Puppeteer to drive a real Chrome instance, so Claude can open a URL, record a performance trace, read console and network activity, inspect the rendered DOM, and automate interactions — then wait for the results and report them. It is aimed at coding agents, but for anyone chasing search problems it is a gift: real browser data instead of guesswork.

Why it’s a must-have for search issues

Google ranks the rendered page, not your source. That is exactly where most SEO issues hide: JavaScript that delays content, images that tank Largest Contentful Paint, a console error that stops a script, a redirect chain that wastes crawl budget. With this MCP, Claude can load your page in real Chrome and see those things directly — the same signals Google’s own systems react to. Instead of 'try compressing images,' you get 'your LCP element is a 1.8 MB hero loading at 4.1s — here’s the fix.'

Diagnosing "why isn’t my page ranking"

Point Claude at a page and ask it to check why it underperforms. Through Chrome DevTools MCP it can run a performance trace and report Core Web Vitals, flag render-blocking resources, list console errors, confirm whether your content is even present in the rendered DOM (a common JavaScript-SEO trap), and check that your structured data and meta tags survive rendering. In minutes you have a prioritized, evidence-backed list — the difference between a hunch and a diagnosis.

From diagnosis to fix in one flow

The real power shows when you pair it with tools that can act. Claude measures the problem in Chrome, then — with your file or CMS connected — makes the change and re-measures to prove it worked. That loop (observe, fix, verify) is how good performance work actually happens, compressed into one conversation. On this very site, that pattern is how we hold Lighthouse 100 and sub-second loads: measure in a real browser, change, confirm the number moved.

What it will not do (and why that’s fine)

Chrome DevTools MCP is a measurement and automation tool, not a ranking oracle. It will not tell you which keyword to target or write your content strategy. What it removes is the technical fog: it turns 'the site feels slow' and 'Google isn’t indexing us' into specific, reproducible numbers and errors you can act on. Pair it with real content and solid information architecture, and you’ve covered both halves of SEO — the technical and the editorial.

Getting started

It supports Google Chrome (and Chrome for Testing). Connect the MCP server to Claude, then hand it a URL and a question: 'measure Core Web Vitals and list what’s hurting them,' or 'load this page and tell me if the main content renders without JavaScript.' Start on a staging copy so you can change and re-test freely. Once you’ve watched it turn a vague ranking worry into a concrete fix list, you won’t debug search issues blind again.

Sources & further reading

Frequently asked questions

What is Chrome DevTools MCP?

It is Google’s official MCP server that gives an AI assistant like Claude access to Chrome DevTools. It drives a real Chrome via Puppeteer to measure performance, read the console and network, inspect the rendered DOM, and automate browser actions.

How does it help with SEO?

Google ranks the rendered page, and this MCP lets Claude see that rendered page in a real browser — Core Web Vitals, render-blocking resources, console errors, and whether your content and structured data actually appear. It turns vague ranking worries into specific, fixable issues.

Does it replace tools like Lighthouse or Search Console?

No — it complements them. It gives an AI hands-on, real-time access to the same browser diagnostics, so you can move from measuring a problem to fixing and re-measuring it in one flow, rather than reading a static report.

Which browsers does it support?

It officially supports Google Chrome and Chrome for Testing, which it automates with Puppeteer.

Is it hard to set up?

No. You connect the MCP server to Claude, then give it a URL and a question. Working on a staging copy lets you change and re-test freely.

Want a site that passes the real-browser test?

I build hand-coded sites that hit Lighthouse 100 and sub-second loads — measured in real Chrome, not promised on a slide. One flat fee.

Get my free quote

Canva MCP: Turn Claude Into Your On-Demand Design Team

Built2WinWeb — Fri, 10 Jul 2026 00:00:00 +0000

Canva MCP: Turn Claude Into Your On-Demand Design Team

🎨 Canva & MCP · Updated July 2026 · 8 min read · by Jacob Campbell

The bottleneck in most marketing is not ideas — it is production. You know you need a social graphic, a one-pager, a set of ad variations, but each one means opening Canva, finding the brand kit, and nudging boxes for twenty minutes. Canva’s official MCP server removes that friction. It connects Claude to Canva’s design APIs, so you can create and edit designs, pull from your brand assets, search your library, and export finished files by describing what you want. It is the difference between having design software and having a design team on call.

Key facts

Official — A remote MCP server built by Canva
Brand-aware — Uses your brand kit, assets, and library
Create+edit — Generate new designs or edit existing ones
Export — Produce finished, ready-to-share files

What the Canva MCP does

Canva provides an official remote MCP server that exposes Canva’s capabilities as tools an AI assistant can call. Through it, Claude can create and edit designs, manage assets and your brand kit, search your library, export files, and even leave comments. You connect it once, and from then on design work becomes a conversation: describe the piece, the format, and the brand, and watch it get built inside your Canva account rather than starting from a blank artboard.

Why it matters for small teams

If you are a small business or a solo marketer, you are the design team — and that is exactly where hours disappear. The Canva MCP lets you offload the mechanical part while keeping the taste. Ask for 'an Instagram post announcing our summer sale, on-brand, three variations,' and you get drafts to react to in seconds. You still choose what ships; you have just skipped the tedious assembly. For a business trying to stay consistent across channels, that is a real multiplier.

Staying on-brand automatically

The feature that makes this genuinely useful is brand awareness. Because the MCP can reach your brand kit and assets, the output uses your colors, fonts, and logos instead of generic templates. That is the hardest thing to keep consistent when you are moving fast — and the easiest to get wrong when a human is rushing. Handing the constraints to the tool ('use our brand kit, our logo, our palette') means every draft starts already looking like you.

A workflow you can run today

Try a batch job: 'Create five ad variations for our new service — same offer, different headlines and layouts, all on brand — then export them as PNGs.' Claude drafts them in Canva; you open the ones you like, tweak, and publish. Or go the other way: point it at an existing design and ask it to resize for a new platform or refresh the copy. The pattern is always the same — you set the brief and the brand, the MCP does the production, and you approve.

Where a human still decides

Canva MCP speeds up making; it does not replace judgment about what to make. Knowing which message lands, when a layout feels cluttered, and whether a campaign fits your voice — that is still you. The healthy way to use it is as a fast first-drafter: generate options quickly, then apply the taste only a person brings. Used that way it does not flatten your brand into AI sameness; it frees you to spend your attention on the decisions that matter.

Getting connected

Canva runs a remote MCP server, so setup is mostly authorizing it to your account through Canva’s AI connector, then pointing Claude at it. Start with low-stakes assets — a social post, an internal one-pager — to get a feel for how it interprets briefs. Give it clear constraints (format, brand, tone) and it delivers usable drafts; leave it vague and it guesses. As with any tool that touches your brand, review before you publish.

Sources & further reading

Frequently asked questions

What is the Canva MCP server?

It is Canva’s official remote MCP server that lets AI assistants like Claude use Canva’s design capabilities — creating and editing designs, managing brand assets, searching your library, exporting files, and commenting — by describing what you want.

Will designs match my brand?

Yes. Because the MCP can access your brand kit and assets, it can build with your colors, fonts, and logos rather than generic templates — as long as you tell it to use your brand.

Does it replace a designer?

No. It accelerates production — first drafts, variations, resizes — while you keep the judgment about what to make and what ships. Think of it as a fast drafter, not a creative director.

How do I set it up?

Canva runs a remote MCP server, so you authorize it to your Canva account through Canva’s AI connector and point Claude at it. Start with low-stakes assets to learn how it interprets briefs.

Is it safe to let it edit my designs?

You control which account it connects to and can start with new or duplicate designs. As with any tool touching your brand, review the output before publishing.

Need a site as sharp as your new visuals?

I build fast, custom websites that make your brand look the part — and set up the AI workflows that keep your content flowing. One flat fee.

Get my free quote

Higgsfield MCP: Generate Video & Images From Claude

Built2WinWeb — Fri, 10 Jul 2026 00:00:00 +0000

Higgsfield MCP: Generate Video & Images Straight From Claude

🎬 Higgsfield & MCP · Updated July 2026 · 8 min read · by Jacob Campbell

Every business now needs video and images — for ads, social, product shots, explainers — and most cannot staff a media team to make them. Higgsfield’s MCP closes that gap in an unusual way: it gives Claude direct access to a large library of generation models, so you can create images and videos, train characters, and browse your history without leaving the conversation. Instead of learning five different AI tools, you describe what you want and let one agent pick the right model and produce it.

Key facts

30+ — Image and video models available through one MCP
Auto-pick — Your agent chooses the best model — or you specify
Characters — Train and reuse consistent characters
One session — Generate, iterate, and browse history in one place

What Higgsfield MCP is

Higgsfield uses MCP — the open standard that gives AI agents access to external tools — to expose its generation platform to assistants like Claude. Once connected, your agent can generate images, create videos, train characters, and browse your creation history, all within a single session. It reaches 30-plus models (Soul, Cinema Studio, Flux, Seedream, Kling, Minimax Hailuo, Veo and more), and it can pick the best model for the task automatically or use the one you name.

Why one MCP beats five tabs

The current reality of AI media is fragmentation: one tool is best for realistic video, another for stylized images, another for character consistency — each with its own login, credits, and quirks. Higgsfield’s MCP collapses that into a single interface. You state the goal ('a 6-second product teaser, cinematic, our brand colors') and the agent routes it to the right model. For a small team, not having to learn and juggle a dozen generators is most of the value.

Consistent characters and brand assets

The feature that turns novelty into a workflow is character training. Generic AI media looks generic partly because nothing carries over between shots. Being able to train a character — a mascot, a spokesperson, a product — and reuse it means your video and images can stay recognizably yours across a campaign. Combined with history you can revisit, it starts to feel less like a slot machine and more like a studio you can iterate in.

A practical marketing loop

Here is a realistic use: ask Claude, through Higgsfield’s MCP, to generate three image concepts for a launch, pick one, then produce a short video built around it, and finally a set of resized cuts for different platforms. You iterate by conversation — 'make the second one warmer, tighten the ending' — rather than re-learning each tool’s controls. Pair it with a design MCP for layout and you have most of a content pipeline driven from one place.

Keeping quality and rights in mind

Generation is fast, which makes taste and judgment the scarce inputs, not effort. Review outputs the way you would a junior creative’s — for brand fit, accuracy, and tone — before anything ships. Be mindful of usage rights and platform policies for AI-generated media, especially in ads, and keep a human in the loop on anything customer-facing. Used deliberately, it is a force multiplier; used carelessly, it just makes more mediocre content faster.

Getting started

Connect Higgsfield’s MCP to Claude, then start small: generate a few images from a clear brief and see how the model selection feels. Add video once you trust the image output, and try training a character if consistency matters to your brand. The more specific your brief — subject, style, length, mood — the better the result. Once you have produced a usable clip from a single sentence, the workflow sells itself.

Sources & further reading

Frequently asked questions

What is Higgsfield MCP?

It is an MCP connection to Higgsfield’s generation platform. Once connected, an AI agent like Claude can generate images and videos, train characters, and browse your creation history in one session, with access to 30+ models.

Do I have to choose which model to use?

No — the agent can automatically select the best model for the task, or you can specify one (for example Kling or Veo for video). That is a big part of the appeal: you focus on the brief, not the tool.

Can I keep a consistent look across a campaign?

Yes. You can train and reuse characters, which helps video and images stay recognizably yours across shots — the thing generic AI media usually lacks.

Is AI-generated media safe to use in ads?

Review outputs for brand fit and accuracy, keep a human in the loop on customer-facing content, and check usage rights and platform policies for AI-generated media before publishing.

How do I start?

Connect Higgsfield’s MCP to Claude and begin with a few images from a clear, specific brief. Add video and character training once you trust the output. Specific briefs — subject, style, length, mood — produce the best results.

Have the content — need somewhere to show it off?

I build fast, custom websites that make your video and visuals shine, plus the AI workflows to keep them fresh. One flat fee.

Get my free quote

Website Redesign vs Rebuild: Which Do You Need?

Built2WinWeb — Fri, 10 Jul 2026 00:00:00 +0000

Website Redesign vs Rebuild: How to Know Which You Need

🔧 Redesign or Rebuild · Updated July 2026 · 8 min read · by Jacob Campbell

You know your website needs help. What is not obvious is whether you need a redesign — a new look on the same foundation — or a rebuild, where the whole thing is reconstructed properly underneath. Get this wrong and you waste money: a beautiful redesign on a slow, broken foundation is lipstick on a problem, while a full rebuild of a site that only needed a refresh is overkill. Here is how to tell the difference before you pay for either.

Key facts

Look — A redesign changes how the site looks and reads
Foundation — A rebuild fixes how the site is actually built
Speed — The clearest signal you need a rebuild, not paint
Own it — A rebuild is the chance to finally own your site

What each one actually means

A redesign keeps the underlying site and changes the surface: layout, colors, images, copy, and structure. It is the right call when the foundation is sound but the site looks tired or converts poorly. A rebuild reconstructs the site from the ground up — new, clean code underneath. It is the right call when the foundation itself is the problem: too slow to fix, too bloated to optimize, or built on a platform you cannot escape. One changes the paint; the other replaces the frame.

Signs you only need a redesign

If your site already loads fast, works well on mobile, and ranks reasonably — but simply looks dated or does not guide visitors to act — a redesign is likely enough. The bones are good; you need better presentation. Clearer calls to action, updated visuals, sharper copy, and a layout aimed at conversion can lift results without touching the foundation. Do not pay for a rebuild you do not need. If the only real complaint is 'it looks old and doesn’t convert,' start with design.

Signs you need a full rebuild

Some problems no redesign can fix. If your site is slow no matter what you try, built on a heavy page builder stacked with plugins, impossible to make mobile-friendly, invisible in search, or on a platform where you pay monthly and own nothing — those are foundation problems. A prettier front end on that base still loads slowly and still ranks poorly. When speed, SEO, or ownership is the issue, you need new code underneath, not a new coat of paint.

The speed test settles most cases

When in doubt, look at load time. Speed comes from the foundation — how the site is coded and what it is built on — so a slow site almost always means a rebuild, because you cannot design your way to fast. If your site loads quickly and just looks off, redesign. If it drags and no amount of image-shrinking or caching fixes it, that is the foundation talking, and a rebuild is the honest answer. Speed is the single most reliable signal of which project you actually need.

Why a rebuild often pays off

A redesign makes a site prettier; a rebuild makes it faster, easier to rank, and finally yours. Done right, a rebuild replaces a slow, rented, plugin-heavy site with a hand-coded one that loads in under a second, is built for search, and comes with no monthly fees. If you were going to invest in the look anyway, and the foundation is weak, doing both at once — new design on new code — usually costs less than redesigning twice and delivers a site that actually performs.

Don’t lose your rankings in the process

Whichever you choose, protect what is working. A careless rebuild can tank your search rankings by breaking URLs and losing content Google already trusts. A proper rebuild preserves your SEO — mapping old URLs to new with redirects, keeping your content and metadata, and improving speed on top. That is the difference between a rebuild that grows your traffic and one that resets it. Ask anyone you hire exactly how they will protect your rankings; the good ones have a clear answer.

Not sure which you need?

The quickest way to know is to have someone look at how your site is actually built, not just how it looks. I will tell you honestly whether a redesign will do the job or whether the foundation needs replacing — and I will not sell you a rebuild you do not need. When a rebuild is the answer, I deliver a fast, hand-coded, SEO-safe site you own outright for one flat fee. Get a free quote and I will give you a straight answer.

Frequently asked questions

What’s the difference between a redesign and a rebuild?

A redesign changes how the site looks and reads while keeping the same foundation. A rebuild reconstructs the site with new, clean code underneath. Redesign fixes presentation; rebuild fixes how the site is actually built.

How do I know which one I need?

Start with speed. If your site loads fast and just looks dated, a redesign is usually enough. If it’s slow no matter what, built on a heavy platform, or you pay monthly and own nothing, those are foundation problems that need a rebuild.

Will a rebuild hurt my Google rankings?

Only if it’s done carelessly. A proper rebuild preserves SEO by mapping old URLs to new with redirects and keeping your content and metadata, while improving speed — so rankings grow rather than reset.

Is a rebuild worth the extra cost?

Often, yes. A redesign makes a weak site prettier but still slow; a rebuild makes it fast, rankable, and owned by you. If the foundation is weak and you’re investing in the look anyway, doing both at once usually costs less than redesigning twice.

Can’t I just make my current site faster?

Sometimes, if it’s well built. But sites on heavy page builders and plugins often can’t be made truly fast without new code underneath — which is a rebuild, not a redesign.

Redesign or rebuild? Get a straight answer.

I’ll tell you honestly which your site needs — and if it’s a rebuild, deliver a fast, hand-coded, SEO-safe site you own for one flat fee.

Get my free quote

The Backlink Gap: Why Competitors Outrank You

Built2WinWeb — Fri, 10 Jul 2026 00:00:00 +0000

The Backlink Gap: Why Competitors Outrank You (and How to Close It)

🔗 Backlink Gap · Updated July 2026 · 9 min read · by Jacob Campbell

You have a faster site and better content, and a competitor still ranks above you. Frustrating — and usually explainable in one word: backlinks. Search engines treat links from other sites as votes of trust, and the business with more quality votes tends to win. The gap between their links and yours is the backlink gap, and it is one of the most actionable things in SEO, because it hands you a precise to-do list: the exact sites already linking to your competitor that you can go earn links from too.

Key facts

Votes — Backlinks act as trust signals to search engines
The gap — Sites linking to competitors but not to you
A list — Gap analysis turns SEO into a concrete to-do
Reachable — Those sites already link in your niche — pitch them

Why backlinks still decide rankings

For all the changes in SEO, links remain one of the strongest signals of authority. When a reputable site links to yours, it is vouching for you, and search engines factor that trust into who ranks. This is why two similar businesses with similar sites can rank very differently — one has earned more quality links than the other. You can do everything else right and still lose to a competitor with a stronger link profile. That is not unfair; it is just a signal you have not competed on yet.

What a backlink gap actually is

A backlink gap is precise: it is the set of domains that link to your competitor but not to you. That definition is powerful because it is a finished list of targets — not 'go get more links' in the abstract, but 'here are the specific sites already willing to link to a business like yours.' They link in your niche, they linked to your rival, and they have not linked to you yet. Close that gap and you are not inventing an outreach strategy; you are completing one your competitor already proved works.

How to find the gap

Finding it is straightforward with the right tool: you enter your site and one or more competitors, and the tool returns every domain linking to them that is missing for you. Historically this meant an expensive Ahrefs or Semrush subscription, but that is no longer the only option — there are now free tools built on open web-crawl data that run gap analysis without a monthly bill. The output is a ranked list of real, reachable sites. That list is the single most useful artifact in a link-building campaign.

Turn the list into links

A list of target sites is only valuable if you act on it. For each one, figure out why they linked to your competitor — a resource page, a roundup, a guest post, a mention — and offer them a reason to link to you too: better content, an updated resource, a genuine contribution. Pull the contact, send a specific, useful pitch (not a template blast), and follow up. Most links come from a modest number of good, personalized asks. The gap tells you who to ask; the outreach earns the yes.

Quality beats quantity

Do not chase every link indiscriminately. A handful of relevant, reputable links outweighs hundreds of junk ones, and low-quality link schemes can actively hurt you. The beauty of gap analysis is that it points at sites already linking within your niche — inherently relevant, inherently more valuable. Prioritize the reputable, on-topic domains from the list, earn those honestly, and ignore the spammy shortcuts. Steady, quality links from sites that matter is how you durably close the gap and climb.

It’s a repeatable advantage

The best part: this is not a one-time trick. Competitors keep earning new links, so the gap keeps refreshing — which means you can re-run the analysis regularly and always have a current list of proven targets. Businesses that treat backlink gap analysis as an ongoing habit, not a one-off, steadily build authority while rivals who ignore it plateau. It turns the vague, intimidating world of link building into a simple loop: find the gap, earn the links, repeat.

Close the gap that’s holding you back

If a competitor keeps outranking you despite your better site, the backlink gap is likely why — and it is fixable with a clear, repeatable process. I build fast, SEO-ready sites and can help you map the exact links your competitors have that you don’t, then go earn them. Get a free quote and let’s turn their advantage into your target list.

Frequently asked questions

What is a backlink gap?

It’s the set of domains that link to your competitor but not to you. It’s valuable because it’s a finished list of sites already willing to link to a business like yours — the exact targets for earning links that could lift your rankings.

Why do backlinks matter for ranking?

Search engines treat links from other sites as trust signals. A business with more quality links tends to rank higher, which is why two similar sites can rank very differently — one has earned more authority.

How do I find my backlink gap?

Use a gap-analysis tool: enter your site and competitors, and it returns every domain linking to them but not you. This once required an expensive subscription, but free tools built on open web-crawl data can now do it.

How do I actually get those links?

For each target, understand why they linked to your competitor, then offer a genuine reason to link to you — better content, an updated resource, a real contribution — and send a specific, personalized pitch. Quality outreach beats template blasts.

Is chasing lots of backlinks a good idea?

No. A few relevant, reputable links outweigh hundreds of junk ones, and spammy schemes can hurt you. Gap analysis helps because it points at sites already linking in your niche — inherently relevant and worth earning honestly.

Turn your competitors’ links into your target list

I build SEO-ready sites and help you find and earn the backlinks competitors have that you don’t. One flat fee, built to climb the rankings.

Get my free quote

Custom MCPs: Give Your Company Its Own AI Superpowers

Built2WinWeb — Fri, 10 Jul 2026 00:00:00 +0000

Custom MCPs: Give Your Company Its Own AI Superpowers

🏢 Custom MCP · Updated July 2026 · 10 min read · by Jacob Campbell

Public MCP servers — Chrome, Canva, GitHub, Higgsfield — are impressive, but they all do generic things. The real unlock for a business is the MCP nobody else has: one that wraps your systems — your inventory, your booking engine, your pricing rules, your internal database — so an AI like Claude can act inside your actual operation. MCP is an open standard, which means you are not waiting for a vendor to build it. You (or someone like me) can build a custom MCP for your company, and turn every internal tool into something your AI can safely use.

Key facts

Open — MCP is an open standard — anyone can build a server
Your data — Wrap internal tools, databases, and APIs
Any client — One server works with Claude and other MCP apps
Own it — No vendor lock-in — you control what it exposes

Why generic tools hit a ceiling

Off-the-shelf MCP servers are broad by design — they know Canva or GitHub, not your business. The moment a task touches something proprietary — 'check stock in our warehouse system,' 'book this into our scheduler,' 'apply our pricing tiers' — a public server cannot help, because it has never heard of your systems. That gap is where most of a company’s real work lives. A custom MCP is how you close it: you teach the AI your business, once, in a way it can act on every day.

What a custom MCP actually is

An MCP server is a small program that exposes a set of tools an AI can call — 'look up an order,' 'create an invoice,' 'query sales for last week.' Because MCP is an open standard, you can wrap almost anything behind those tools: a REST API, a database, a spreadsheet, an internal admin panel. You define what the AI is allowed to do and how, and the same server works with Claude and any other MCP-compatible client. It is less 'build an AI' and more 'give your existing systems a clean, safe door the AI can knock on.'

What companies can build

The practical list is long: a support agent that reads your order and returns systems to answer customers accurately; an ops assistant that queries your database and drafts the weekly report; a sales tool that checks live inventory and applies your pricing; an internal helper that files tickets, updates records, or kicks off a workflow. Each of these is a handful of well-defined tools wrapped around systems you already run. The AI does not need to understand your whole business — just the specific, safe actions you expose.

Safety and control are the point

Building your own server is also how you keep control. You decide exactly which actions exist, what data they can touch, and where the guardrails sit — read-only where it should be read-only, confirmation required for anything destructive, sensitive fields never exposed. Because you own the server, nothing leaves your control that you did not deliberately allow. That is a very different posture from pasting company data into a chat box, and it is why custom MCPs are how serious businesses adopt AI without giving up security.

You don’t need a huge team to start

This is more approachable than it sounds. A first custom MCP can be tiny — two or three tools around your single most-used system — and still change how your team works. Start with the task you do a hundred times a week, wrap just that, and connect it to Claude. Prove the value, then add tools. Anthropic and the MCP community provide the SDKs and patterns; the hard part is not the code, it is deciding clearly what the AI should be allowed to do.

How to get one built

You have two paths: build it in-house if you have developers who can spend a focused week on it, or have someone build it for you as a defined project. Either way, the recipe is the same — pick a real workflow, define the tools and their guardrails, wrap the underlying system, and connect it. This is exactly the kind of work I take on alongside building sites: a custom MCP that makes your specific business faster, owned by you, with no monthly platform holding it hostage.

Sources & further reading

Frequently asked questions

What is a custom MCP server?

It is a small program you build that exposes your own tools and data — an internal API, a database, a scheduler — as actions an AI like Claude can safely call. Because MCP is an open standard, you control exactly what it does.

Why build one instead of using public MCP servers?

Public servers are generic — they know Canva or GitHub, not your inventory, booking system, or pricing rules. A custom MCP wraps your proprietary systems so the AI can act inside your actual business, which is where most real work lives.

Is it secure to connect AI to our internal systems this way?

Done right, it is more secure than pasting data into a chat. You define which actions exist, keep sensitive things read-only or off-limits, require confirmation for destructive actions, and own the server — so nothing is exposed that you did not deliberately allow.

Do we need a big engineering team?

No. A first server can wrap just two or three tools around your most-used system and still be transformative. Start small, prove value, and expand. The SDKs and patterns are provided by Anthropic and the MCP community.

Can you build one for my company?

Yes — it is exactly the kind of project I take on: pick a real workflow, define the tools and guardrails, wrap your system, and connect it to Claude. You own it, with no monthly platform fee.

Want a custom MCP built for your business?

I build custom websites and the AI tooling around them — including bespoke MCP servers that wire your own systems into Claude. One flat fee, fully owned by you.

Get my free quote

Crawlgraph MCP: Automate Backlink Research in Claude

Built2WinWeb — Fri, 10 Jul 2026 00:00:00 +0000

Crawlgraph MCP: Automate Competitor Backlink Research in Claude

🔗 Crawlgraph & MCP · Updated July 2026 · 8 min read · by Jacob Campbell

Backlink research is powerful and, let us be honest, tedious — export a list, sort it, cross-reference competitors, dig up contacts. It is exactly the kind of repetitive, structured work that AI should handle. Crawlgraph’s MCP server makes that real: it connects Crawlgraph’s backlink lookups and competitor gap analysis to Claude, so you can just ask — 'find the sites linking to these three competitors but not to me' — and the agent runs the tools and hands you the answer. It turns a spreadsheet grind into a conversation.

Key facts

In Claude — Run gap analysis by asking, no dashboard
Any client — Works with Claude, Cursor, Cline, and other MCP apps
Plain English — Describe the goal — the agent runs the tools
Gap + links — Backlink lookups and competitor gap analysis

What Crawlgraph MCP does

Crawlgraph’s MCP server exposes its backlink lookups and competitor gap analysis to AI assistants through the Model Context Protocol. Once connected, you describe what you want in plain language and Claude runs the underlying Crawlgraph tools for you — no dashboard, no manual exports. It works not just with Claude but with any MCP-compatible client like Cursor or Cline. In short, it takes the backlink data Crawlgraph already provides and makes it something you can query by conversation rather than by clicking.

Why running it through Claude changes things

The value is not just convenience — it is workflow. On its own, a gap-analysis tool gives you raw data you then have to interpret and act on. Through the MCP, Claude can run the analysis and immediately help with the next steps: summarize the most promising targets, group them by why they linked, draft outreach for the best ones. The research and the action collapse into one session. You go from 'who links to my competitors?' to 'here are ten to pitch and here are the emails' without leaving the conversation.

A real research loop, in one conversation

Picture the flow: you ask Claude to run a Crawlgraph gap analysis on you versus three competitors. It returns the domains linking to them but not you. You ask it to filter for the most relevant and reputable, and to explain why each likely linked. Then you ask it to draft a short, specific pitch for the top five. What used to be an afternoon of exports, sorting, and copy-pasting is now a ten-minute back-and-forth. The MCP does not just fetch data; it lets you reason and act on it in place.

Cheap data meets cheap automation

What makes this combination genuinely useful for small businesses is cost. Crawlgraph already offers the core gap analysis for free on open Common Crawl data, without the monthly subscription of the big tools. Add the MCP and you get automation on top of affordable data — the two expensive parts of backlink research, both handled cheaply. You are not paying enterprise prices for the data or building brittle scripts for the automation; you connect a server and ask. That is a real leveling of a field that used to favor whoever could afford Ahrefs.

Getting set up

Setup follows the usual MCP pattern: connect Crawlgraph’s MCP server to Claude (or your preferred MCP client), then start asking. Begin with a simple request — a gap analysis against your top competitor — to see how it interprets the task, then layer on filtering, prioritizing, and drafting. As with any tool that pulls real data, sanity-check the results and treat the outreach drafts as starting points to personalize. The point is speed on the tedious parts, with your judgment still steering the campaign.

Where it fits

Crawlgraph MCP is one piece of a larger idea I keep coming back to: connect the real tools your business needs to an AI that can operate them, and the repetitive work evaporates. Backlink research is a perfect candidate — structured, repeatable, and valuable. Pair it with a fast, well-built site and genuine outreach, and you have an SEO loop that a solo business can actually sustain. The MCP does not replace strategy; it removes the drudgery that used to keep small businesses from doing this at all.

Let AI do your backlink grind

If backlink research has always felt too tedious or too expensive to keep up, running Crawlgraph through Claude changes both. I build fast, SEO-ready sites and set up AI workflows — including MCP-driven research like this — so your link building runs on autopilot for the boring parts. Get a free quote and let’s wire up a site and a strategy that outwork your competitors.

Frequently asked questions

What is the Crawlgraph MCP server?

It connects Crawlgraph’s backlink lookups and competitor gap analysis to AI assistants through the Model Context Protocol. You describe what you want in plain language and Claude runs the Crawlgraph tools for you — no dashboard or manual exports.

Which AI tools does it work with?

Any MCP-compatible client — Claude, Cursor, Cline, and others. You connect the server once and run backlink and gap analysis by asking.

How is this better than using Crawlgraph directly?

It collapses research and action into one session. Claude can run the gap analysis, then summarize the best targets, explain why they linked, and draft outreach — turning raw data into next steps without leaving the conversation.

Is it expensive?

No. Crawlgraph offers the core gap analysis free on open Common Crawl data, and the MCP adds automation on top — so both the data and the automation, usually the pricey parts of backlink research, stay affordable.

How do I get started?

Connect Crawlgraph’s MCP server to Claude or your MCP client, then start with a simple gap analysis against your top competitor. Layer on filtering, prioritizing, and drafting, and sanity-check the results as you go.

Automate the backlink grind

I build fast, SEO-ready sites and set up AI workflows like Crawlgraph’s MCP so your link research runs itself. One flat fee, built to outrank.

Get my free quote

DEV Community: Built2WinWeb

Keeping a Website Healthy: What Maintenance Covers, Who Should Do It

Key takeaways

I Made My Site Fully Agent-Readable. Here Are the Real Numbers.

I Made My Site Fully Agent-Readable. Here Are the Real Numbers.

By the numbers

What I actually built

What happened: the numbers

Cloudflare Crawler Hints was firing 4,900 URLs every 18 hours

My CDN was contradicting my own robots.txt

Discovery and citation are different problems

The bug that was hiding underneath all of it

What I would tell you to do

So was it worth doing?

If you want a site built this way

Related reading

Frequently asked questions

Did adding MCP, A2A and llms.txt improve search rankings?

What is Cloudflare Crawler Hints and should I turn it off?

Why would my CDN change my robots.txt?

Is IndexNow worth setting up?

What should a small site do first?

Want a site machines can actually read?

Turn Your Website Into a Lead-Generating Machine

How to Turn Your Website Into a Lead-Generating Machine

Key facts

It gets found in the first place

It loads before they lose patience

It makes the next step obvious

It earns trust fast

It captures the lead cleanly

It’s built to keep working

Want a website that actually sells?

Related reading

Frequently asked questions

Why isn’t my website generating any leads?

What’s the single most important factor?

How many calls to action should a page have?

Do I need reviews to generate leads?

Can my current website be turned into a lead machine?

Turn your website into a lead machine

OBS Studio MCP: Run Flawless Class Streams Without the Panic

OBS Studio MCP: Run Flawless Class Streams Without the Panic

Key facts

What OBS MCP does

Why this fits teachers so well

Learn the mental model, not the menus

Set up a class recording in minutes

Staying in control during a live class

The payoff for your students

Sources & further reading

Related reading

Frequently asked questions

What is OBS MCP?

What do I need to run it?

Do I need to be technical?

How does it help teachers specifically?

Is it safe?

Want a home for your lessons and recordings?

Roblox Studio MCP: Learn Game Design Before You Teach It

Roblox Studio MCP: Learn Game Design Before You Teach It

Key facts

What the Roblox Studio MCP is

Why teachers are the perfect users

Watch the Luau, learn the logic

A starter project for your class

Keeping it classroom-safe

From confident teacher to confident class

Sources & further reading

Related reading

Frequently asked questions

Is there an official Roblox MCP server?

Do I need to know Luau or Studio first?

What does it cost?

How does this help me teach?

Is it safe for a classroom?

Running a game-design or coding program?

Chrome DevTools MCP: The Must-Have for Search & SEO Issues

Chrome DevTools MCP: The Must-Have for Search & SEO Issues

Key facts