The argument in one line.
Skills are not better prompts -- they are pre-wired tool pipelines that offload the decision tree from the model to the file itself, which is why they run faster, cost less compute, and produce consistent output across an entire team.
Read if. Skip if.
READ IF YOU ARE…
- You run an SEO team or agency and write the same research prompts from scratch each week.
- You are already using Claude Desktop and want real ranking and analytics data plugged in without building custom scripts.
- You need to produce and distribute client SEO reports without manually stitching together Data Studio, GA4, and a spreadsheet.
- You want to understand what agentic SEO looks like before committing to a workflow change.
SKIP IF…
- You do not use SE Ranking -- the skills are open source but the data layer requires an SE Ranking account to produce real output.
- You are looking for a prompt engineering tutorial; this is a product demo for a specific toolchain.
The full version, fast.
SE Ranking built and open-sourced 24 Claude skills across four SEO categories: content, competitive research, AI search visibility, and strategy. Each skill is a markdown file that tells Claude exactly which SE Ranking MCP tools to call, in what sequence, and how to format the output, so the model spends zero tokens deciding what to do. The headline demo cuts content brief production from seven hours to ninety seconds plus one hour of review. A master SEO plan skill sits on top, auto-detecting which sub-skills to run and synthesizing a roadmap from real data. The video closes with a live artifact -- a Claude-built HTML dashboard connected to SE Ranking and GA4 that refreshes on demand without additional credit spend.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Where the time goes.
00:00 – 00:43
01 · Why We Built 24 SEO Skills
Hook and promise: 7 hours to 1 hour for a content brief, single-prompt client reports. Skills are the repeatable mechanism.
00:43 – 02:04
02 · What Claude Skills Are
A skill is a markdown file that specifies tool calls, order, and output format. Lighter on compute than verbose prompts because the decision path is pre-encoded.
02:04 – 02:38
03 · Installing the SEO Skills Library
One-step install via Claude Desktop plugins marketplace: seranking/seo-skills. 24 skills available in every chat after sync.
02:38 – 05:10
04 · SEO Content Brief Skill Demo
7 sequential steps: competitor detection, keyword gap, topic cluster selection, SERP analysis, top-article reading, internal link audit, duplicate page check. Everything lands in one file.
05:10 – 05:48
05 · The SEO Plan Skill Demo
The orchestrator skill auto-selects sub-skills, runs them, and builds a roadmap from real data. No prompt assembly required.
05:48 – 08:12
06 · Building Live Artifacts with Claude
Claude builds a live HTML dashboard from SE Ranking MCP + GA4 simultaneously. Build costs ~18% of Claude Pro credit block; refreshes are free. Exportable as PDF.
08:12 – 08:57
07 · The Future of SEO Workflows
Outro: in 1-2 years SEO is MCPs + skill libraries + agents running automatically. No dashboard logins.
Lines worth screenshotting.
- A Claude skill is a markdown file that hardwires which tools to call and in what order -- the model executes, it does not decide.
- Skills run lighter on compute than verbose prompts because token spend on decision-making is eliminated at the file level.
- The content brief skill finds your actual competitors by shared keyword overlap, not your perceived competitors.
- Before writing a single heading, the skill checks if you already have a page targeting that keyword and tells you to update it instead of creating a duplicate.
- The SEO plan skill is an orchestrator that calls other skills as sub-routines, building a full strategic roadmap from real data.
- Building a live artifact costs roughly 18% of your Claude Pro 5-hour credit limit; every subsequent refresh is free.
- Live artifacts pull SE Ranking and GA4 data simultaneously, replacing Data Studio, N8N, and Make for client reporting.
- The 24 skills span four areas -- content, competitive research, AI search visibility, and strategy -- each mapping to a task someone was prompting from scratch every time.
- The hardest problem with AI search data has been freshness and accuracy; MCP-connected live queries solve both without a custom pipeline.
- An SEO team can install the entire 24-skill library in one sync and have it available in every Claude Desktop chat immediately.
One skill file beats a hundred saved prompts
WHAT TO LEARN
Encoding the decision tree into the file rather than the prompt is what separates a repeatable workflow from a folder of prompts you still have to adapt every time.
- A skill specifies which tools to call, in what order, and what to do with the output -- the model executes rather than decides, which is why results are consistent across a team.
- Filtering keyword opportunities inside the skill (informational intent, 1k+ monthly searches, KD under 40) means those constraints are never forgotten or misapplied.
- Checking for an existing page before generating a new content brief prevents content cannibalization -- it only works reliably when baked into the skill, not left to the user to remember.
- An orchestrating skill that calls sub-skills as subroutines produces a coordinated output from real data; without it, the same work requires running multiple separate prompts and assembling the results yourself.
- A live artifact built from MCP-connected data refreshes on demand for free after the initial build -- the expensive step is the build, not the maintenance.
- Splitting skills into four clear categories makes the library navigable and eliminates the which-prompt-do-I-use overhead that accumulates in a flat folder.
Terms worth knowing.
- Claude skill
- A markdown file that instructs Claude which MCP tools to call, in what order, and how to format the output for a specific repeatable task -- eliminating the need to re-describe the workflow each time.
- MCP (Model Context Protocol)
- A standard that lets Claude call external data sources and tools directly inside a conversation, here used to pull live SE Ranking ranking, keyword, and competitor data.
- Live artifact
- A Claude-built HTML dashboard connected to live MCP data sources that can be refreshed on demand without rebuilding or re-prompting, and exported as a PDF for clients.
- Keyword gap
- Keywords your competitors rank for that your site does not -- filtered here to informational intent, over 1,000 monthly searches, and keyword difficulty under 40.
- AI search visibility
- How often and prominently a brand appears in responses from AI assistants (ChatGPT, Perplexity, Gemini, AI Overviews) when users ask relevant queries.
- Orchestrator skill
- A top-level skill that determines which other skills are needed for a given request, runs them as sub-routines, and assembles their outputs into a unified result.
Things they pointed at.
06:33toolGA4 MCP (Google Analytics 4)
Lines you could clip.
00:52
“Skills are actually lighter on compute than long, verbose prompts because Claude isn't spending tokens figuring out what you want. It already knows what to do.”
02:59
“Not who you think you're competing with, who you actually are.”
05:41
“Every recommendation you're looking at in this roadmap came from real data, not Claude inventing something that sounds reasonable.”
07:52
“Building the artifact costs credits, but refreshing it doesn't. You can hit that button as many times as you want without touching your credit balance.”
08:38
“In a year or two, you're not going to log in into a dashboard to do SEO. You'll have a stack of MCPs, a library of skills, and agents running workflows automatically.”
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
analogy
00:00We built 24 clog skills to replace the most tedious SEO tasks eating up our team's time. One of these skills cut our content brief creation from seven hours to just one. Another builds a refreshable client report with actual ranking and analytics data from a single prompt.
00:20They're all open source, need a couple of clicks to install, and built for every SEO task we used to write a new prompt for every single time. In a few minutes, you'll see these run live and what the workflow is behind them.
00:34Then we'll connect Claude to GA four and build something that means you never have to write dozens of prompts again, ever. Let's go. A skill is basically a markdown file.
00:46It tells Claude exactly how to approach a repeatable task, as in which tools to call, in what order, and what to do with the output. Here's what's worth knowing.
00:57Skills are actually lighter on compute than long, verbose prompts because Claude isn't spending tokens figuring out what you want. It already knows what to do.
01:07Let's take a closer look at one of these skills. If say, you want the content brief to always pull your specific competitors, or always output any format your team actually uses, you just go in and change it.
01:21That's the whole point of making these open source. They're a starting point, not a black box. Now, we didn't just build one skill.
01:29We built 24 skills our SEO team uses regularly. They sit across four areas, content, competitive research, AI search visibility, and strategy.
01:40Each one handles something you'd normally be writing a prompt for from scratch every single time. But here's what makes them different from just a collection of prompts. They're connected.
01:50There's one called SEO plan that sits on top of everything else. When you run it, it figures out which other skills it needs, runs them, and builds the output from what comes back. We'll show you how it works in a few minutes.
02:04Now, you can install the whole library in one step. Open your Cloud desktop version, go to customize, personal plugins, create plugins, add marketplace.
02:16Type s two ranking forward slash s e o hyphen skills, hit sync, and there they are, sitting there ready in every chat you open. The skills themselves are free, but they run on s two ranking data, so you'll need to connect your account.
02:32If you don't have one yet, there's a link in the description. Now, let me show you a couple of these running. We'll start with the content brief.
02:39FYI, the last brief we built manually took us seven hours, but this one takes around ninety seconds to run and about an hour to review. Let me show you why the gap is that big.
02:51When you run this skill, it does seven things in sequence. Things you'd normally do across seven different tabs. It finds your top five competitors by shared keyword overlap.
03:02Not who you think you're competing with, who you actually are. From there, it runs a keyword gap. Keywords your competitors rank for that you don't.
03:10Filtered automatically to informational intent, over a thousand searches a month. Difficulty under 40.
03:17It picks the best topic cluster from that gap and tells you why it picked it. Once it has a topic, it goes deeper, full SERP analysis, AI overviews, people also ask, related keywords, plus it then actually reads the top three ranking articles and pulls the signals you'd need to match to compete.
03:37Then it looks at your own site, finds the pages you already have that could link to this new article, and tells you exactly where to put the anchors. It's important to note here that before it writes a single heading, it checks whether you already have a page targeting this keyword. If you do, it tells you to update it instead of creating something new.
03:58That's fewer bad decisions for you. Everything lands in a single file a writer can open and start from. Let's run it.
04:06Type forward slash SEO hyphen content hyphen brief. Specify the domain, topic, and country. Here's what comes back.
04:15Primary keyword, volume, difficulty, and intent.
04:19Three title options, a meta description, then the full heading structure from h one down to h three.
04:26Each section has what to cover and what to link out to. This part is worth looking at. These are the specific things the top three ranking articles don't cover.
04:36That's where you beat them. Existing pages on your site, the anchor text to use, and which section of the new article they belong in. You don't have to figure that out yourself.
04:47Then we have the brands LLMs currently cite for this topic, and the specific things you'd need to do to get cited alongside them. Seven hours down to one.
04:57The hour that's left is reviewing this and handing it to a writer or a Claude agent. By the way, we're working on a fully automated version of our AI content creation workflow. Subscribe so you don't miss it.
05:10Now, remember the SEO plan from earlier? We said it calls the other skills. Let me show you what that looks like.
05:16Type forward slash SEO hyphen plan and specify the domain. You can see the skills firing one after another. It's figuring out what it needs, runs whatever skills are necessary, and then builds everything from what came back.
05:31I wanna point out that every recommendation you're looking at in this roadmap came from real data, not Claude inventing something that sounds reasonable. Before we wrap up, I have one more thing to show you, and it's not a skill. It's a little something called live artifacts.
05:48A live artifact is basically a dashboard that Claude builds for you from real data and then keeps up to date. You don't run the skill again. You hit refresh whenever you want updated numbers, and you can export it as a PDF for a client.
06:03Now, we're gonna use this prompt, which by the way, you can copy from the page we linked in the description. But let's open it with the phrase, build me a live artifact. We're using SC Rankings MCP, which is a two click install, and GA four at the same time here.
06:20If you don't have GA four connected yet, there's a link in the description to Google's setup guide. Or you can check out this video where we show you how we connected these MCPs step by step. Back to our video.
06:32Let's watch this thing get built live. So here's what we've got. A search visibility across Chad GPT, Perplexity, Gemini, AI overviews.
06:41Share a voice against your top competitors and GA four sessions from AI referral sources over the last ninety days compared against the ninety days before that. Very cool.
06:52Now, the hardest part with AI search data has always been quality. Is it real? Is it fresh?
06:58This will no longer be an issue. This setup is pulling from IstaRankings live queries and your actual GA four account.
07:06This also means you don't need Data Studio, N8N, or Make anymore. You can forget about spending long afternoons figuring out how to connect things, which can be a real nightmare.
07:17Now here's the part I genuinely like most about this. This button, hit it, and everything updates with the latest data. I no longer need to create a new prompt to ask Claude to update it.
07:28By the way, building the artifact costs credits, but refreshing it doesn't. You can hit that button as many times as you want without touching your credit balance.
07:38The expensive part is the build. One build, roughly 18% of your five hour clock pro limit.
07:46After that, refreshes are free. Another thing, SC ranking and GA four data are talking to each other inside one dashboard, but you're not limited to one data source.
07:57You can connect as many MCPs as you need and build from all of them at once. If you wanna see which ones are worth adding, that's exactly what we cover in our best MCP servers video. The link's in the description.
08:10Last thing, promise. If you need to share this with a client or your team, export it as a PDF. They get the same data without the login and cloud account needed.
08:21Just a clean report they can open. That's the full stack. This is exactly what we meant at the start of this video.
08:28In a year or two, you're not going to log in into a dashboard to do SEO. You'll have a stack of MCPs, a library of skills, and agents.
08:38Agents running workflows automatically. That's where this is going. Now, with the skills installed, your next question should be, which MCPs are actually worth connecting to them?
08:48We've tested the ones that move the needle for SEO. Watch that next in this video.
The Hook
The bait, then the rug-pull.
Seven hours to build a content brief. That was the baseline before SE Ranking's marketing team started encoding their workflows into Claude skills -- markdown files that tell Claude exactly which tools to call and what to do with the output, with no re-prompting, no tab-switching, and no forgetting to filter by intent.
Named ideas worth stealing.
02:38list
The 7-Step Content Brief Pipeline
- Find top 5 actual competitors by keyword overlap
- Run keyword gap (informational intent, 1k+ searches/mo, KD < 40)
- Pick best topic cluster with justification
- Full SERP analysis (AI Overviews, PAA, related keywords)
- Read top 3 ranking articles for gap signals
- Internal link audit with exact anchor placements
- Duplicate page check before generating headings
The sequence the /seo-content-brief skill runs automatically in a single prompt, replacing seven separate manual research steps.
Steal forAny content team that currently builds briefs by hand across multiple tabs
01:35list
Four SEO Skill Categories
- Content
- Competitive Research
- AI Search Visibility
- Strategy
The four domains SE Ranking's 24 skills cover, each mapping to a class of tasks previously done via one-off prompts.
Steal forOrganizing any team's AI skill library by workflow category rather than by tool
How they asked for the click.
VERBAL ASK
05:09subscribe
“Subscribe so you don't miss it”
Soft mid-video subscribe ask tied to a future automated content creation workflow -- well-timed after the brief demo payoff
MENTIONED ON CAMERA
FROM THE DESCRIPTION
Visual structure at a glance.
