Claude Code for SEO and GEO: Audits and Reports Without Coding

How to use Claude Code for SEO and GEO without coding: what it costs, how to install it, which audits and reports it automates, what skills are, how to connect Search Console through MCP and how to handle client data.
Written with AI, reviewed by humans

In this article

See how your brand shows up in AI answers

Book a demo and find out what to fix first, with a GEO specialist.

Book a demo

Reading time: 10 minutes

Claude Code helps with SEO and GEO as an assistant that reads files, opens pages and runs tasks from plain-language requests. With it, you can audit AI crawler access, join exports and draft reports without coding. As for Claude Code pricing, it starts with the Pro plan at $20 per month.

In short

  • Claude Code comes with every paid Claude plan, starting with Pro. The Free plan does not include it.
  • You don’t need to code, because the desktop app skips the terminal and you make requests in plain language.
  • In SEO and GEO, it works best on tasks over files and public pages, such as robots.txt, crawls and Search Console exports.
  • Before you use client data, check the plan’s training policy and block access to passwords and keys.

What is Claude Code and how much does it cost?

Claude Code is the Anthropic agent that reads files, edits them, runs commands and connects to other tools. It works in the terminal, in code editors, in a desktop app and on the web. So, to use Claude Code for SEO, the minimum plan is Pro, at $20 per month on monthly billing.

On Anthropic’s pricing page, Claude Code pricing appears in dollars and without tax, because it comes inside the Claude plans:

PlanPublished priceClaude Code
Free$0Not included
Pro$20 per month or $17 on annual billingIncluded
MaxFrom $100 per monthIncluded, with more usage
Team$25 per standard seat or $20 on annual billingIncluded

However, Claude Code pricing does not buy unlimited use. According to the same page, terminal work and chats draw from one usage pool, which resets in five-hour windows. Paid plans also add weekly limits. Therefore, a big audit in the morning can leave less room for chat in the afternoon.

For an agency, then, Claude Code pricing depends on who uses it. A solo analyst starts on Pro. In contrast, a team that will open client data pays Claude Code pricing on Team, per seat, for the reasons in the last section.

How do you install Claude Code?

In general, the simplest path is the Claude desktop app for Windows or macOS. According to the Claude Code documentation, you install it, sign in and click the Code tab. The app already includes Claude Code, but it requires a paid plan.

If you prefer the terminal, the docs give one command per system. In Windows PowerShell, for example:

irm https://claude.ai/install.ps1 | iex

On macOS and Linux:

curl -fsSL https://claude.ai/install.sh | bash

Then open a new terminal and run claude --version. If the install worked, the command prints a version number. The system requirements call for macOS 13 or newer, or Windows 10 version 1809 or newer, plus 4 GB of memory.

Claude Code for SEO: illustration of a code window producing report cards with line, bar and gauge charts

Do you need to know how to code to use Claude Code?

No. With Claude Code for SEO, you describe the task in plain language. Then the tool plans the steps, reads the files in the folder and asks for permission before sensitive actions. Still, it helps to know the basics of folders, CSV files and robots.txt, because you check the result. Anthropic itself writes in its terminal guide for beginners:

“You can use Claude Code even if you’ve never used a terminal before.”

Claude Code documentation, Anthropic (2026)

In practice, the work happens in one folder on your computer. For example, you put the crawler export, the Search Console CSV and the prompt list there. Next, you ask for the analysis as you would ask an analyst. Besides text and CSV files, Claude Code reads images and PDFs, including screenshots of reports.

Even so, the review stays with you. After all, the model can get a sum wrong or read the wrong column, and the client will get that number. To compare Claude Code with chat tools, n8n and agents, see the guide to AI for SEO and GEO.

Which SEO and GEO audits can you automate with Claude Code?

Claude Code for SEO works best on repeated checks over public pages and exported files. That includes the robots.txt of many sites, joining a crawl with Search Console data and reviewing metadata. In other words, it does the legwork, and you decide what goes into the report.

  • AI crawlers: read each domain’s robots.txt and list who blocks GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot.
  • Crawling: join the crawler export with the Search Console export and find pages with impressions and errors.
  • Metadata: flag empty, duplicate or overly long titles and descriptions in a spreadsheet.
  • Content: list pages with no date, no author or no internal links to the main topic.

What to check in each item, and why, is in the technical SEO audit checklist. Claude Code for SEO and GEO does not replace that judgment, but it speeds up data gathering.

Example: AI crawlers in the robots.txt of three sites

This is the kind of task you can hand to Claude Code: “Read the robots.txt of nytimes.com, en.wikipedia.org and washingtonpost.com and say whether GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot can access the pages.” For the table below, we did the check by hand instead. On November 1, 2026, we read the three public files and compared the rules for each crawler. The sites are a major U.S. newspaper, the English-language Wikipedia and another U.S. newspaper. These files change often, so read them again before you rely on the table.

SiteCrawlers namedResult
nytimes.comAll fourBlocked site-wide
washingtonpost.comClaudeBot and PerplexityBotThose two blocked site-wide; the other two follow the general rules
en.wikipedia.orgNone of the fourAllowed by the general rules

In fact, at The New York Times, each crawler has a “Disallow: /” line, and at The Washington Post, ClaudeBot and PerplexityBot do. However, robots.txt is only part of the answer, since the CDN and the firewall also block crawlers. That is why you should read which AI crawlers to allow and which to block before you change a client’s file.

How do you build reports and monitoring with Claude Code?

Use Claude Code to read exports and draft the report, not to collect the answers. Through the Claude in Chrome extension, Claude Code for GEO can open pages, but in a visible window and in real time. So, collecting at volume calls for each engine’s API, paid separately, or a GEO tool.

First of all, volume matters. The protocol for measuring mentions and citations calls for about 1,000 answers per client per month. Beyond volume, the API can cite different sources from the ones a consumer sees in the app, as the GEO tools comparison shows. Once the answer sheet is ready, Claude Code calculates presence and citation rates and writes the monthly draft. To show the numbers to the client, an SEO dashboard in Data Studio, formerly Looker Studio, combines that sheet with GA4 and Search Console.

To repeat the task every month, Claude Code for GEO has two official paths:

  • Scheduled task on desktop: it runs on your computer, with access to local files. But according to the scheduled tasks documentation, it only fires while the app is open and the computer is awake.
  • Cloud routine: it runs even with the computer off, on the Pro, Max, Team and Enterprise plans. However, routines are in research preview and work on a fresh copy of a repository, without local files.

When the report changes, it helps to understand why the AI answer changes. See why ChatGPT cites one brand and not another. For a fixed flow that runs the same way every month, compare it with the guide to n8n for SEO and GEO.

What are SEO and GEO skills for Claude Code?

In short, a skill is a SKILL.md file with instructions that Claude Code follows when the task calls for it, or when you type /skill-name. In SEO, it stores a repeated procedure, such as the AI crawler checklist or the monthly report format. As a result, every analyst runs the same audit the same way.

According to the skills documentation, a personal skill lives in ~/.claude/skills/name/SKILL.md and works across all your projects. A project skill, on the other hand, lives in .claude/skills/name/SKILL.md, inside the work folder, and the whole team can use it. Skills also follow the Agent Skills open standard, which works across other AI tools.

People who search for “claude code seo skill” find ready-made skills on GitHub, built by third parties. But before you install one, read the whole file. According to the docs, a skill can grant itself tools without asking for permission each time, through the allowed-tools field. Therefore, a third-party skill deserves the same review as any program downloaded from the internet.

How do you connect Search Console to Claude Code through MCP?

MCP is an open standard that links Claude Code to other SEO tools, and each service needs an MCP server. The Search Console servers we found come from third parties, built on Google’s official API. For that reason, the safest way to start is still to export the Search Console CSV into the project folder.

If you choose MCP, the MCP documentation shows the command for a remote server:

claude mcp add --transport http <name> <url>

At the same time, Anthropic warns that you should trust each server before you connect it. This is because servers that fetch outside content carry a risk of prompt injection. For GA4, on the other hand, Google itself maintains a Google Analytics MCP server, still experimental and with read-only access.

In the Search Console API, each query returns up to 25,000 rows, and access can be read-only. So, prefer a server that asks only for read permission. That way, it cannot submit sitemaps or change the client’s property.

How do you protect client data in Claude Code?

Check three points before you open client data: the plan, retention and permissions. On the Free, Pro and Max plans, you choose whether your chats train models. On Team, Enterprise and the API, by contrast, Anthropic does not train on the data by default. Therefore, for client data, Claude Code pricing on Team pays off.

According to the Claude Code data policy, retention also changes with the plan. On personal plans, it is 30 days without training and 5 years with training on. On commercial plans, however, the default is 30 days. In addition, your own computer keeps chats in plain text for 30 days, in the ~/.claude/projects folder.

In practice, follow this order:

  1. Set up one folder per client and open Claude Code only in it.
  2. Block sensitive files with a deny rule in the permissions, such as Read(./.env), which stops it from reading the keys file.
  3. Give read-only access to Search Console, GA4 and any MCP server.
  4. Check every number before you send the report to the client.

What else you can hand off to an agent, and what still needs approval, is in the guide to AI agents for SEO.

Want to know how your brand shows up in AI answers today?

Book a demo and see how your brand shows up in AI answers, with support from a GEO specialist.

Main sources

About this articleWritten with AI and reviewed by humans, with verified sources. Found something out of date? Write to the NotHuman team.

Get new NotHuman articles by email

By subscribing, you agree to receive blog articles by email and to the Privacy Policy. Unsubscribe any time.

Keep reading

Demo

Want to know how your brand shows up in AI answers today?

Book a demo and see how your brand shows up in AI answers, with support from a GEO specialist.

Book a demo

Your privacy

We use necessary cookies to make the site work and Google Analytics and Microsoft Clarity statistics cookies to improve our content. You can decline the statistics ones; in the EU, UK and Switzerland they are only turned on after you accept. Cookie Policy.

Cookie preferences

Cookie Policy