Claude SEO sub-skills organized around the central SEO orchestrator.
Capturas do projetoSobre este Skill
Claude SEO is an open-source Claude Code plugin for website audits and search optimization. It routes requests to focused workflows for technical SEO, content, schema, sitemaps and AI search, and can combine findings into a prioritized action plan. Start with a single page when the question concerns one URL; use a full audit when the scope is the site.
Por que escolhemos
Useful for SEO consultants, agencies, and website teams who want audits and follow-up work in their agent workflow. This listing links to Agrici Daniel's public MIT-licensed project. Both diagrams are the author's assets; optional data services and model usage have separate costs.
No que ajuda
- SEO consultants and agencies reviewing client sites and organizing follow-up work.
- SaaS, publisher, and e-commerce teams checking technical issues and content quality.
- Website owners checking schema, sitemaps, AI-search readiness, and page experience.
- Local and international businesses reviewing location signals or language targeting.
What you can do
Website and page audits
Check crawlability, indexing, metadata, internal links, page experience, images, and sitemaps. Full audits coordinate specialist agents and combine their findings.
Content and structured data
Review content quality and E-E-A-T signals, prepare content briefs, and detect, validate, or generate Schema.org markup.
AI search and agent readiness
Review how clearly a page answers questions and supports attribution. Agent readiness workflows inspect access, Markdown, llms.txt, and WebMCP where relevant.
Business-specific analysis
Use workflows for local SEO, backlinks, e-commerce, international targeting, keyword clustering, competitor pages, and SEO change monitoring.
Official repository and READMEOfficial command referenceSEO Skill instructions
Install the official plugin
The author also provides inspected-repository installation scripts for macOS, Linux, and Windows. Follow the official installation guide for those paths and for updates.
Add the marketplace
Run this command inside Claude Code to register the author's public repository.
/plugin marketplace add AgriciDaniel/claude-seoInstall Claude SEO
Install the plugin with its specialist Skills, agents, and supporting runtime files.
/plugin install claude-seo@agricidaniel-claude-seoSet up and check the runtime
Run setup once to provision the isolated Python environment and Chromium. Use doctor to check readiness afterward.
/seo setup /seo doctor
Start with a focused task
For a first Claude Code SEO check, start with one important URL. After installing the plugin, replace the example URL below, run the page command and give Claude the audience, purpose and question you want answered.
| Comando | Para que serve |
|---|---|
/seo page https://example.com/about | Analyze one page in detail. |
/seo audit https://example.com | Audit a complete website and prioritize findings. |
/seo schema https://example.com | Check and generate structured data. |
/seo geo https://example.com | Review AI-search readiness and page content. |
/seo technical https://example.com | Focus on technical SEO issues. |
Run your first page audit
This suggested brief helps keep the review bounded. Replace the placeholders and send it after the page command; it is an example request, not a promised result.
Give the page context
Explain who the page serves and what you need to decide. Ask the agent to separate what it can observe from checks that need other data.
Review [URL] for [audience]. The page aims to [goal]. My question is [question]. List the evidence and URL for each finding. Separate confirmed observations from assumptions and checks requiring Search Console or another service. Give me three prioritized actions and a validation step for each. Do not change the site.
What happened when we audited a real page
On November 4, 2026, SkillHey ran Claude SEO 2.5.0 (commit 6be93f4) with Claude Code 2.1.306 and Claude Opus 5.5 against our Best Video Skills article. The session was restricted to the single page and did not use Search Console, a browser renderer, shell tools or subagents.
The tool reported that the article had no visible byline or publication date. We confirmed those were absent from the visible article, while the raw HTML already contained SkillHey as author and the publication date in Article structured data. We added a visible byline and date to the article; this is an editorial improvement, not evidence of a ranking change.
The tool received a Markdown conversion of the page, so it could not inspect the raw head, canonical, robots or structured data. It suggested checking them; we separately verified that the English page returned index, follow and a self-referencing canonical. Its scores are subjective and were not validated against search outcomes.
Method and deliverables
The illustrated method follows four phases: perceive, analyze, validate, and act. Recommendations connect observed evidence, dependencies, a way to check failure, and a leading indicator. The main output is a Markdown report with findings and a prioritized plan.
Selected workflows also provide structured data, sitemaps, or machine-readable results. The Google reporting workflow documents PDF reports. Available outputs depend on the chosen command, configured runtime, and data access.
Optional data and costs
Start with publicly reachable pages. Add Google credentials when you need PageSpeed, CrUX, Search Console, or GA4 data. Optional MCP extensions can supply crawling, keyword metrics, backlink intelligence, AI visibility, or image generation; install only the integrations your task needs.
The public repository is free under MIT. A paid community mirror is separate from this listing. Agent/model usage and optional service plans are not included with the open-source plugin.
Choose the right SEO workflow
Use the single-page workflow for a specific URL and the site audit for broader crawl and technical questions. Supply the target keyword and intended audience so the recommendations address the actual search intent.
A page audit can identify visible indexing instructions and content gaps. Search Console impressions, rankings and click data require access to the relevant property; without it, treat traffic conclusions as unverified.
Antes de usar
- Use Claude Code with local file and terminal access. The official plugin installation requires Claude Code 1.0.33 or newer.
- The bundled Python tools require Python 3.11 to 3.14 with pip. Run /seo setup to provision the managed runtime.
- Playwright Chromium is needed for rendered SPA analysis and screenshots. Target websites must be reachable; blocked or interaction-dependent content may need manual review.
- Core workflows do not require paid SEO APIs. Google integrations and optional MCP extensions need their own credentials; Claude Code, model usage, and third-party services may charge separately.
Agentes compatíveis
Perguntas frequentes
Is Claude SEO free?
The public MIT-licensed plugin is free to install. Claude Code/model usage and optional API services can have separate costs.
Which agent should I use?
This repository's documented installation and runtime target Claude Code. Use the author's installation guide for the complete plugin.
Do I need API keys?
Core page analysis works without paid SEO APIs. Google data and optional MCP providers require the corresponding account credentials.
Does it guarantee rankings or AI citations?
No. It identifies issues and suggests work to verify. Search rankings and AI citations depend on factors outside the plugin; review recommendations against your site's facts and goals.
Is Claude SEO a Claude Code SEO checker?
Yes. This open-source Claude Code plugin provides website and page SEO analysis workflows. Review its findings against the available page evidence; optional live data needs separate access or integrations.
How does Claude SEO compare with Marketing Skills SEO Audit?
Claude SEO is a broader plugin with site and page audit workflows alongside schema, AI search and other SEO tasks. Marketing Skills SEO Audit is a focused technical and on-page review entry from a separate project.
Fontes e referências
As fontes originais estão nos links abaixo. As sugestões práticas são recomendações editoriais do SkillHey.
