How to Do SEO and AEO with Claude Cowork (Full Workflow)

Claude Cowork lets you run an entire SEO and AEO pipeline — from detailed keyword research to finished content — in about 20 minutes. Not with generic AI chat. With skills, connectors, and multi-step workflows that execute directly on your machine.

This changes how SEO and AEO work gets done. You're not copying and pasting between tools. You're running complete workflows that chain together research, analysis, and content creation in a single session.

Here's the exact process.

If you've heard of Claude Code, Cowork is the non-technical version. It's a clean UI built on top of Claude Code.

That distinction matters. You're not using regular Claude. You're using Claude Code in the backend — which means it can do far more than a standard chat interface.

It runs directly on your computer. That means it can do almost anything you can do on your machine:

• Create and edit files
• Search the web
• Complete multi-chain workflows
• Access external tools via connectors

Anthropic has done significant safety work on it, so it's broadly secure for most tasks.

1. Skills

Skills are specific capabilities you give Cowork to execute a particular task.

But here's what makes them different from just a prompt — a skill isn't just a markdown file explaining a workflow. It can also package in data and attachments that give it additional context. That means you can create repeatable workflows with built-in reference material.

Examples of skills:

• X content creation
• LinkedIn post writing
• Blog and case study generation
• UI design
• SEO keyword research
• Content writing with brand voice

We're going to use two skills in this workflow: one for SEO keyword research and one for blog creation.

2. Connectors (MCP Apps)

Connectors are how Cowork talks to external applications. These are MCP apps — the same standard that works across ChatGPT, Claude, Cursor, and VS Code.

Build one connector and it works everywhere. So if you have an MCP app, you can deploy it in the ChatGPT App Store, in Cursor, in VS Code — all using the same standard.

3. Subagents

Claude Cowork can delegate to specialized subagents that execute tasks in their own context window. This saves context and keeps each task focused.

Maybe it spins up a subagent specifically to do coding. Another to do research. Each one works independently without fighting for context space.

4. Commands

Commands let you create repeatable workflows. Maybe every day you want a daily update on keyword changes — you create a command for that. Trigger it and it runs the entire workflow, using the right skills, connectors, and subagents automatically.

That's how the whole thing gets brought together.

First, leverage as many MCP apps as possible — they're native and integrated.

Available connectors:

• Ahrefs MCP connector
• Google Cloud BigQuery (connects to Google Analytics & Search Console data)
• Firecrawl MCP (for scraping websites)
• Web search (built-in)

Don't have Ahrefs or SEMrush? No problem. They're expensive and not everyone needs a full subscription. You can use CSV exports instead. Download your keyword data from Ahrefs (even on a basic plan), export from Google Analytics and Google Search Console, and upload those directly.

In Cowork, go to Settings → Connectors to see what you have. Click Browse to explore available connectors and add what you need.

Pro tip: Even without premium SEO tool subscriptions, CSV uploads work perfectly. The AI doesn't care whether the data came from an API or a spreadsheet.

Skills are where the quality comes from. We're using two:

Skill 1: Keyword Research & Strategy

This skill follows a structured methodology:

• Phase 1: Context gathering — understanding your business, competitors, and current performance
• Phase 2: Data integration and analysis — processing Ahrefs/SEMrush exports and Google Analytics data
• Phase 3: Strategic analysis — identifying keyword opportunities, gaps, and priorities
• Two deliverables: a strategy document and a content action roadmap

The skill includes a methodology file that defines exactly how to approach keyword research — volume distribution, difficulty analysis, CPC highlights, growth trends, and keyword clusters.

Skill 2: SEO Content Creation

This skill takes the output from the keyword research skill and creates content:

• Picks articles based on the content roadmap priorities
• Writes SEO and AEO-optimized content following best practices
• Generates metadata, FAQ sections, and internal link suggestions
• Also includes a methodology file for consistent quality

Both skills will be available as downloads — drop a comment if you'd like them.

To install: go to Settings → Capabilities → Customize to see all your skills. Add the skill files and they'll be available in every session.

Phase 1: Generate Keyword Ideas from Your Data

Start by getting Cowork to analyze your existing data and suggest keywords. Here's the prompt:

I have here my Google Analytics and Google Search Console data.

What I now want you to do is review all of that data and then have a think on the kind of keywords — based on your understanding of the business — what kind of keywords we could be going after.

I'm also going to attach some more context on what we do and the company. You can use Firecrawl MCP to scrape our website. Here are also some attachments of other documents that are related so you can understand more about the business.

Use that to come up with keyword suggestions around things that we can focus on for our keyword data that we want to get from Ahrefs.

Attach your Google Analytics export, Search Console data, and any company context documents.

Cowork comes back with a structured list of keyword opportunities based on your actual performance data and business context. Take those keywords and pull the data from Ahrefs (or SEMrush, or Data for SEO).

Important: Create a dedicated folder for this work. Keep all your Ahrefs exports, competitor data, and backlink reports in one place. Mine is simply called 'SEO and AEO'.

Phase 2: Run the Full Strategy + Content Pipeline

Once you have your keyword data exported from Ahrefs, this is where the magic happens. One prompt kicks off the entire pipeline:

I now want you to leverage this data that we have from Ahrefs and your understanding of what we do at Ghost Team AI.

I want you to use the keyword research skill to, based on this data, do some research to think about the best SEO and AEO strategy for us, and come up with an action plan as to what to execute.

Then I want you to create the first five articles and blogs on the back of the research that you've done using the SEO Content Creation skill.

Then show me those outputs and I'll review them.

Attach your Ahrefs exports: website overview, competitor performance, competitor backlinks, your backlinks, and keyword data.

That's it. One prompt. Cowork handles the rest.

When you hit go, Cowork creates a task map. Here's the sequence:

• 1. Reviews all Ahrefs data files
• 2. Runs SERP analysis using web search
• 3. Builds keyword strategy and scoring
• 4. Creates the content action roadmap
• 5. Writes all five articles
• 6. Verifies outputs

It picks up both skills automatically — the keyword research skill first, then the SEO content creation skill. You can watch it tick through each component in real time.

It also uses web search to supplement the data. If you had the Ahrefs connector directly, you'd see those API calls here too.

Strategy Documents

First, you get the keyword strategy. This includes:

• Volume distribution across keywords
• Difficulty distribution analysis
• CPC highlights
• Key growth trends
• Keyword clusters grouped by theme

This kind of analysis takes hours manually. Cowork does it in minutes.

Then you get the content action roadmap — a prioritized top 20 list of articles to write, with reasoning for each one.

Content Strategy Insights

The AI understands SEO deeply. In our demo it identified:

• Agentic commerce — massive keyword growth, worth targeting early
• ChatGPT app development — low difficulty, low volume, but zero credible competition. Pure opportunity.
• An internal link blueprint showing how articles should connect
• An AEO checklist for optimizing content for AI consumption

It creates a blend of quick wins (low difficulty) and long-term plays (high growth) automatically.

The Articles

Each article comes with:

• Full long-form content — structured with headings, sections, and logical flow
• SEO metadata tags
• FAQ sections
• Markdown, HTML, and a brief for review

The articles are detailed. You can adjust length — tell it to go shorter or longer. But the default output is substantial and well-structured.

Once the first five are done, you can keep going:

Create the rest of the top 20 pieces from the content roadmap.

Set it running in the background. Come back to finished articles.

This isn't possible in regular Claude. You'd hit context window limitations, lose track of the data, and get generic outputs.

Cowork solves this because:

• Skills keep the methodology consistent across every piece of content
• Subagents handle different phases without fighting for context space
• Connectors pull live data from your tools
• It runs on your computer with access to all your files

The traditional workflow: open a keyword tool, export data, analyze in a spreadsheet, brief a writer, wait for drafts, review, optimize, publish. Multiple tools, multiple people, multiple days.

With Cowork: one session, two prompts, skills handle the quality. Twenty minutes.

What's next: You can integrate this directly into your CMS so content goes from research to draft-ready in your website automatically. That's a workflow we'll cover in a future guide.

Book a strategy call with our team to understand what Claude Cowork and MCP Apps can mean for your business.