Claude SEO is a free, open-source SEO audit plugin for Claude Code that checks a website's technical health, content authority and AI-search visibility, then hands back one ranked list of fixes. It is aimed at the person responsible for a site's organic traffic: an in-house marketer, an agency specialist, or the developer they ask for help, someone comfortable installing a Claude Code plugin and running Python. The repository is MIT licensed, written mostly in Python, and has gathered 23,179 stars and 3,139 forks since it was created in February 2026, with its most recent push in October 2026.
What it does
The premise in the video is narrow and useful: you can rank in Google and still be invisible to AI search. Claude SEO audits both sides of that.
By the project's own description, the coverage is split into 26 sub-skills and 19 specialist sub-agents, spanning:
- technical SEO, crawl errors and Core Web Vitals
- E-E-A-T signals and Schema.org structured data
- GEO and AEO, meaning whether assistants such as ChatGPT, Perplexity, Gemini and Claude can reach and cite the page
- agent readiness: Lighthouse Agentic Browsing, WebMCP and llms.txt
- backlinks, local SEO, e-commerce and international SEO
- Google APIs, plus PDF and Excel reporting
Nine optional extensions pull in live SEO data. The output is not a wall of warnings but a single plan ordered by expected impact, which is the part that makes it worth running over a raw speed report.
How it works
You type one command inside Claude Code. According to the video, up to seventeen agents then work the site in parallel, and ten to fifteen minutes later you get the ranked plan. The sub-skills are the checklists; the sub-agents are the workers that each take one slice and report back, so the breadth comes from fan-out rather than from one long prompt.
Page-level measurement needs a real browser, which is why the project installs its own Chromium instead of driving whatever you already have.
One number travels with this repository: the authors say their own site got the same audit from day one and reached 10.8K clicks and a score of 99 in twelve weeks. That is the authors' claim about their own property, not an independent measurement, and the channel does not run the tools it covers. Read it as a demo, not a forecast. The README also carries badges for CI and for 1,077 passing tests, which at least says the checks themselves are exercised.
Getting started
It needs Claude Code, Python 3.11 through 3.14, and the project's own Chromium install. The Python window matters: interpreters older or newer than that are outside what the video states as supported. Nothing in the material names an operating system, a chip, or a memory floor, so treat hardware as unlisted rather than assuming it runs anywhere.
On money: the plugin itself is free and MIT licensed, but the intelligence is not. Every audit runs on Claude through Claude Code, which is Anthropic's paid product, and seventeen agents working for a quarter of an hour is not a small request. The Google APIs and the nine live-data extensions will want your own accounts and keys; the saved material does not say which ones. The README has Install, Pricing and FAQ sections, but the copy saved here is cut off before them, so take the exact install command from the repository itself: github.com/AgriciDaniel/claude-seo.
When to use it / when not
The main catch is that this is a reporter, not a fixer. It tells you what is wrong and in what order, and you or your CMS still have to do the work, which also means the value depends on having someone able to ship the changes afterwards.
Use it instead of asking Claude to eyeball a page when you want the whole checklist covered in one pass, with the AI-citation questions included rather than bolted on.
Don't use it if you need rank tracking over time, a shared dashboard for a client, or a stored crawl history: this is a point-in-time audit you run yourself, on your own machine.
Alternatives
The tools a reader already knows are the hosted suites, Semrush and Ahrefs for site audits, and Screaming Frog for crawling, plus Google's own Lighthouse and Search Console at the free end. Those give you history and a user interface. Claude SEO gives you breadth instead, including the newer agent-readiness checks that the older suites were not built for, at no licence cost and inside the terminal where you already work.
Take this repository seriously if you own a site's organic traffic, already pay for Claude Code, and want the AI-visibility question answered with the same care as crawl errors and page speed. The scope is unusually wide for a free plugin, the licence is permissive, and the structure of skills plus agents is readable enough to extend with checks of your own. Budget for the model usage, accept that implementing the fixes is still your job, and read the project's own traffic figures as a demo of the workflow rather than a promise about your site.