Run structured SEO audits inside Claude Code with Claude SEO

Claude SEO turns a broad request such as “audit this site” into a routed workflow for technical SEO, content, schema, international targeting, performance, images, and GEO. It is useful when you want repeatable checks and evidence files instead of one unstructured AI answer.

What it does

The official project currently lists 25 sub-skills and 18 specialist agents. A router selects the relevant workflow, while optional integrations can add crawl, search, backlink, or performance data. The result can separate confirmed technical issues from editorial recommendations and claims that still need external evidence.

A sensible first audit

Begin with one page or a small site section. Define the audience, locale, conversion goal, and available data. Ask for findings grouped by severity and require every recommendation to identify its evidence source.

Keep three categories distinct: verifiable defects, judgment-based improvements, and hypotheses that require Search Console, keyword, or competitor data. This prevents plausible AI suggestions from being mistaken for measured SEO facts.

Limits to keep visible

The Skill cannot guarantee rankings or infer private business goals from a public page. Crawling can also be limited by authentication, JavaScript rendering, robots rules, rate limits, and site size. Optional services may add API keys, fees, and separate data-processing terms.

Generated schema or code changes still need validation before production. A clean report is not evidence that the live site has been fixed.

Source and license

Claude SEO is maintained by Agrici Daniel in the official GitHub repository under the MIT License. At review time, the latest GitHub Release is v2.2.4, which includes a managed Python runtime plus sitemap-discovery and security fixes.