How to Run a Full SEO Audit with Claude Code Instead of Eight Tools

Why does an SEO audit take so long?

An SEO audit takes long because no single free tool sees everything. A crawler sees your pages, Search Console sees what Google indexed, a speed test sees loading time and a schema tester sees your markup. You become the glue between them, copying findings into a spreadsheet one row at a time.

Here is what that looks like in practice:

• Many tools, one person. Each tool answers one question. Joining the answers is manual work.
• Steps get skipped. The checklist usually lives in your head. On a busy day, the robots.txt check or the schema test quietly falls off.
• No two audits match. If you run the audit again next month, you probably check things in a different order, so you cannot compare results cleanly.
• The fix list comes last. After hours of checking, you still have to decide what to fix first.

On 2026-09-18 I checked all 1,010 sitemap URLs of my own site isiklisusleme.com. Every one returned 200 with a correct canonical. Yet the og:image tag was broken on 537 pages, and 445 pages had the brand name doubled at the end of the title. A status check alone would have called that site clean.

The checking is not hard. It is long and repetitive, and long repetitive work is where people make mistakes.

How to do an SEO audit by hand, step by step

A manual SEO audit is ten checks done in order, each with a free tool. You can do all of this today without buying anything, and you should do it by hand at least once so you know what a good audit looks like.

1. Crawl your site. A crawler is a program that visits every page like a search engine would. The free version of Screaming Frog SEO Spider crawls up to 500 URLs. Export the list of pages with their status codes.
2. Check what Google indexed. Open the Page indexing report in Google Search Console. "Indexed" means Google stored the page and can show it in results. Note every page listed as not indexed and the reason given.
3. Read robots.txt and the sitemap. Open yoursite.com/robots.txt. Make sure it does not block pages you want found. Then confirm your sitemap lists the pages you care about and nothing that returns an error.
4. Check titles and descriptions. Every page needs one title and one meta description, both unique. Look for missing ones, duplicates and titles that get cut off in results.
5. Check headings. Each page should have one H1, the main headline. Subheadings should follow in order, H2 then H3, like chapters and sections in a book.
6. Find broken links and redirect chains. In your crawl export, filter for 404 errors and for redirects that jump more than once. Fix the link at its source rather than adding another redirect.
7. Test page speed. Run your key pages through PageSpeed Insights. Google's Core Web Vitals call a page good when the main content loads within 2.5 seconds, the page reacts to a tap within 200 milliseconds and the layout shift score stays at 0.1 or lower.
8. Validate schema markup. Schema markup is a hidden label that tells search engines what a page is: a product, an article, a business. Paste each page type into the Rich Results Test and fix every error it reports.
9. Check AI crawler access. AI search tools use their own crawlers, such as OAI-SearchBot, PerplexityBot and ClaudeBot. Make sure robots.txt does not block the ones you want. Our free AI crawler checker does this in one step.
10. Rank the fixes. Put every problem in one list. Score each by impact and effort, then fix the high impact, low effort items first. A missing title on your best-selling page beats a slow page nobody visits.

That last step is the one most audits skip, and it is the one your client or your boss actually reads.

If you want the longer version of which SEO jobs suit automation and which do not, I wrote it up in Claude Code skills for SEO. Google's own SEO Starter Guide is the best free reference for the basics behind each step.

The faster way: run the audit from Claude Code

The faster way is to let Claude Code run the ten checks for you with skills that already contain the checklist. A skill is a saved set of instructions Claude Code loads when a task matches it, a bit like a recipe card it pulls out at the right moment. AI SEO Command Suite is 30 of those recipe cards for SEO work.

I built the pack around one rule from my own SEO work: automate the checks that have a right answer, and keep decisions that need judgement for a person. That is why the audit skills check and verify, and why keyword choices stay with you.

How to use it, step by step

1. Buy and download. Polar or Gumroad sends the download right after payment.
2. Install the skills. Copy every folder that holds a SKILL.md into ~/.claude/skills, the folder Claude Code reads skills from, and the agent files into ~/.claude/agents. Include the four folders that sit directly inside seo-skills: seo-audit, ai-seo, programmatic-seo and schema-markup. The copy commands in the README skip those four, so if /seo-audit does not show up, that is the reason. There is no setup script.
3. Open Claude Code in a terminal, inside your website project or any folder.
4. Run the audit. Type /seo-audit followed by your site address, for example /seo-audit https://yoursite.com. It takes plain text, not flags. Before it starts, it asks a few short questions about site type, goals, priority keywords and scope, and whether you can share Search Console data.
5. Read the report. You get findings grouped by type with a ranked fix list at the end.
6. Fix and check again. Use /schema-markup or /on-page-seo-auditor to draft fixes, ship them, then run /seo-audit again to confirm the problems are gone.

What the output looks like

The /seo-audit report follows the same shape every time, which is what makes month to month comparison possible. It opens with a short summary and the top issues. Then come technical findings (crawling, indexing, speed), on-page findings and content findings. Each finding lists the issue, its impact, the evidence and the fix. It ends with a ranked action plan. Schema and AI crawler checks go deeper in /schema-markup and /geo-audit.

Here is a real excerpt, translated from Turkish. To be clear about where it comes from: it is from the /seo audit command I run on my own sites, not from this pack's /seo-audit, which I have not yet timed on the same site. The target was askedtheai.com, one of my own sites, on 2026-09-28:

The scan covered 561 pages from the 27 September build. Clean: canonicals all point to www and to the page itself; no noindex page in the sitemap. P1-1: the brand appears twice at the end of the title, "| AskedTheAI | AskedTheAI", on 73 pages. P3: title over 65 characters on 403 pages, description over 165 characters on 348.

Read a report like this the way you would read a junior's work. On 2026-07-02 an audit agent told me 5 pages were missing metadata. When I checked, the real number was 3. Since then I quote the number I verified, never the number the agent gave me.

How much time it saves

Here is what I have measured, and only that. The /seo command read 561 pages of askedtheai.com in 6 minutes 47 seconds on 2026-09-28. On 2026-09-25 it crawled 753 built HTML files of a lighting-decoration site I worked on, set aside 81 redirecting pages and audited the other 522 in 12 minutes 37 seconds. I never timed a manual audit of either site, so I will not claim a ratio. You can time your own: do the ten checks above by hand once, then run /seo-audit on the same site.

The bigger saving is not the first audit. It is the fifth one, when you run the same command and get a report you can put next to last month's and compare line by line.