AgriciDaniel/claude-seo

Claude SEO – an AI‑agent skill for Claude Code

What it is – An open‑source skill (plugin) for Claude Code, the Anthropic‑powered coding assistant. The skill adds a /seo command suite that can audit an entire website (or a single page) and return a structured, prioritized action plan for search‑engine optimisation.

Why it matters – SEO audits are traditionally manual, time‑consuming, and often inconsistent. Claude SEO turns the audit into a repeatable, parallelised AI workflow that:

• Spawns up to 17 specialist agents at once, covering technical SEO, content quality (E‑E‑A‑T), Schema.org markup, AI‑search optimisation, local & e‑commerce SEO, and more.
• Produces recommendations that are falsifiable: each finding includes the observation it rests on, a “how would we know this failed?” check, and a leading indicator.
• Runs locally inside Claude Code’s isolated Python environment, so no data leaves the user’s machine.

---

Core capabilities (selected)

Area	What Claude SEO does
Parallel audits	Up to 17 specialist agents run concurrently; a full‑site audit finishes in minutes.
Technical SEO	Checks Core Web Vitals (LCP, INP, CLS), crawlability, canonical chains, hreflang, etc.
Content & E‑E‑A‑T	Scores experience, expertise, authority, trustworthiness against Google’s Search Quality Rater Guidelines (2025).
Schema.org	Detects, validates, and can generate JSON‑LD for 20+ schema types; warns about deprecated types.
AI‑search optimisation	Aligns with Google’s AI Optimization Guide: passage citability, question‑based heading hierarchy, llms.txt evidence, and AI‑mode readiness.
Local & Maps	Audits Google Business Profile, NAP consistency, reviews, and produces geo‑grid rank tracking.
E‑commerce & International	Checks product schema, marketplace signals, hreflang, and multi‑location doorway‑page limits.
Integrations	Optional connectors to Google APIs (PageSpeed, CrUX, Search Console, GA4, Indexing), Ahrefs, DataForSEO, SERanking, Bing Webmaster, Matomo, and a local Lighthouse runner.
Output	Markdown report + optional PDF (WeasyPrint) + JSON files; all stored locally.

---

Who should use it

• SEO agencies managing multiple client sites – replace quarterly deep audits with fast, weekly automated runs.
• In‑house SEO leads at SaaS, publishing, or e‑commerce firms – get a second pair of eyes before executive reviews.
• Freelance consultants – generate a concrete 0‑100 health score and falsifiable recommendations during discovery calls.

---

Installation (quick start)

1. Via Claude Code marketplace (recommended)

   /plugin marketplace add AgriciDaniel/claude-seo
   /plugin install claude-seo@agricidaniel-claude-seo
   /seo setup
   

   This creates an isolated Python env and installs Playwright Chromium.
2. Manual install (Unix/macOS/Linux) – clone the repo and run install.sh (or the PowerShell script on Windows). The script sets up the same environment without touching global Python packages.

---

Example workflow

# Start Claude Code
claude

# Full site audit (minutes)
/seo audit https://example.com

# Single‑page deep dive
/seo page https://example.com/about

# Schema audit & generation
/seo schema https://example.com

# AI‑search optimisation check
/seo geo https://example.com

# Generate a sitemap from templates
/seo sitemap generate

The commands return a markdown report that lists findings, severity, and a concrete “how to verify” step.

---

Limitations & current scope

• Data sources – Relies on publicly available APIs (Google, Ahrefs, etc.). If API keys are missing, only the free Tier 0 data (PageSpeed & CrUX) is available.
• Local execution only – No cloud service; audits run on the user’s machine, so performance depends on local CPU/RAM.
• Coverage – While 26 sub‑skills are provided, niche SEO factors (e.g., voice‑search schema, emerging Core Web Vitals beyond INP) are not yet implemented.
• Community version – A private mirror (AI‑Marketing‑Hub/claude-seo) offers early‑access features for paid community members; the public repo is the stable, MIT‑licensed baseline.

---

License & contribution

• License: MIT (full source available on GitHub).
• Contributing: Follow the CONTRIBUTING.md guidelines; pull requests are welcomed, especially for new schema types, additional API connectors, or improved falsifiability checks.

---

Bottom line

Claude SEO turns the traditionally manual, expertise‑heavy process of SEO auditing into a fast, reproducible AI‑driven workflow that lives entirely on the user’s machine. It is especially useful for agencies and consultants who need frequent, data‑rich audits without recurring SaaS fees, and for anyone already using Claude Code as their AI development environment.