DOCSEARCHbyAlgolia
Submit Docs

Claude SEO

repository·main·Indexed 12 days ago

https://github.com/agricidaniel/claude-seo

An open-source SEO analysis plugin for Claude Code (v2.2.4) that automates technical, content, and AI-search (GEO) audits. It features a system of parallel specialist agents and supports extensions for image generation via Banana, live SEO data via DataForSEO, and full-site crawling via Firecrawl.

Tokens
181.4K
Snippets
355
Records
795
Agent score
95%

What's inside Claude SEO

  1. Browse the FLOW Framework Operational Prompt Library

    main

    The FLOW Framework provides a categorized library of operational prompts designed for different stages of the SEO workflow. You can use these prompts to guide AI agents through specific SEO tasks such as research, optimization, and conversion.

    Prompts are organized into the following stages:

    • find: Used for discovery and research (e.g., Keyword research, Audience avatars, Content planning).
    • leverage: Used to analyze competitive landscapes (e.g., Backlink competition).
    • optimize: Used for refining existing content or creating new assets (e.g., Technical audits, Blog post writing, Schema generation, CTR audits).
    • win: Used for high-intent conversion tasks (e.g., BOFU Page Brief Generator, Conversion audits).
    • local: Specialized prompts for Local SEO and Google Business Profile (GBP) optimization (e.g., GBP categories, GBP descriptions, Local service page rewriting).

  2. Use the seo-local skill for Local SEO analysis

    main

    The seo-local skill is designed for performing comprehensive Local SEO audits. It covers Google Business Profile (GBP) optimization, NAP (Name, Address, Phone) consistency, citation health, review signals, local schema markup, and location page quality.

    It is automatically triggered when a user mentions terms like:

    • "local SEO"
    • "Google Business Profile" or "GBP"
    • "map pack" or "local pack"
    • "citations"
    • "NAP consistency"
    • "service area"
    • "multi-location"

    You can invoke it manually by providing an optional URL as an argument.

    seo-local [url]

  3. Overview of Claude SEO Sub-skills

    main

    The seo skill acts as an orchestrator for 24 specialized sub-skills designed for comprehensive SEO analysis. These sub-skills cover technical, content, local, and strategic SEO tasks. When performing a full site audit, the orchestrator uses parallel sub-agents to execute these tasks simultaneously.

    Core Sub-skills

    • seo-audit: Full website audit with parallel delegation.
    • seo-page: Deep single-page analysis.
    • seo-technical: Technical SEO (crawlability, indexability, security, CWV).
    • seo-content: E-E-A-T and content quality.
    • seo-schema: Schema markup detection and generation.
    • seo-geo: AI Overviews / GEO optimization (llms.txt, citability).
    • seo-local: Local SEO (GBP, NAP, citations).
    • seo-google: Google SEO APIs (GSC, PageSpeed, CrUX, etc.).
    • seo-backlinks: Backlink profile analysis.
    • seo-ecommerce: E-commerce SEO intelligence.
    • seo-flow: FLOW framework integration (Find -> Leverage -> Optimize -> Win).

  4. What is Hub-and-Spoke Content Architecture?

    main
    Hub-and-Spoke architecture is a content clustering strategy designed to build topical authority. It consists of one central pillar page (the hub) that provides broad coverage of a high-volume topic, connected to multiple spoke clusters. Each cluster contains 2-4 individual posts that provide deep-dive coverage into specific subtopics. This structure ensures that the pillar page acts as a central resource while the spokes build depth and internal linking strength.

  5. How the SERP Overlap Methodology works

    main

    The SERP Overlap Methodology is a clustering technique used to determine content architecture based on Google's ranking decisions rather than keyword text similarity.

    Core Principle:

    • Same Page: Keywords that return the same Google results should be targeted by the same page.
    • Separate Pages: Keywords that return completely different results require separate pages.

    This method uses the intersection of organic URL sets to decide whether keywords should be merged into a single post, placed in the same cluster, or treated as entirely separate topics.

  6. Manage GA4 Data API Token Budgeting

    main

    The GA4 Data API uses a token-based quota system rather than simple request counts.

    Token Consumption Estimates:

    • Simple 1-dimension, 1-metric report: ~1-5 tokens
    • Complex multi-dimension, multi-metric: ~10-100 tokens

    Quota Limits:

    • Daily limit: 25,000 tokens per property per project
    • Hourly limit: 5,000 tokens per property per project
    • Concurrency: Maximum 10 simultaneous requests

    Best Practice: Set returnPropertyQuota: true in your requests to monitor token consumption.

  7. How the Managed Python runtime works

    main

    Claude SEO uses a managed Python runtime to ensure tool stability and isolation. Instead of relying on the user's working-directory-relative Python, all bundled tools are dispatched through bin/claude-seo and scripts/runtime.py.

    Key Runtime Characteristics

    • Python Version: Requires Python 3.10 or newer.
    • Operations: The standard-library runtime provides run, setup, and a read-only doctor command.
    • Isolation: Plugin environments are stored under the persistent CLAUDE_PLUGIN_DATA directory. For manual installs, the compatible virtual environment is located at ~/.claude/skills/seo/.venv.
    • Execution: The run operation accepts only allowlisted script basenames or contained extension scripts. It forwards arguments without a shell, preserves child exit codes, and forces UTF-8 child streams.
    • State Management: A state marker tracks the runtime schema, requirements SHA-256, Python version, plugin version, and browser state. Changes to requirements or the Python ABI require an explicit setup call.

  8. How Claude SEO handles Single-Page Applications (SPAs)

    main

    Claude SEO supports modern web frameworks like Next.js, React, and Vue by using a headless renderer.

    By default, the plugin uses an auto-detection mode that identifies SPA hallmarks (such as empty <div id="root"> shells or hydration markers) and switches to a rendered fetch.

    If you need to control the rendering behavior manually, you can use the following scripts:

    • scripts/render_page.py: The shared headless renderer backed by Playwright Chromium. Use --mode auto for auto-detection, or --mode always to force rendering.
    • scripts/fetch_page.py: A wrapper for page fetching. Use --render auto to opt-in to auto-detection, or --render always to force rendering. Note that the default is --render never (raw HTTP).

    Note on Limitations: Pages with scroll-bound hydration or content that only fetches after user interaction may produce noisy findings. For these edge cases, use the seo-visual cross-check workflow.

    # Force rendering via render_page.py
python scripts/render_page.py --mode always

# Force rendering via fetch_page.py
python scripts/fetch_page.py --render always

  9. Choose between seo-unlighthouse and seo-google for audits

    main

    When performing SEO audits, choose the appropriate skill based on your data needs:

    • Use seo-unlighthouse when: You need a site-wide, multi-page audit, want to avoid PageSpeed Insights API quotas, or need to run audits locally/offline.
    • Use seo-google psi or seo-google crux when: You need single-URL field data (CrUX) specifically from Google's datasets.
    • Use scripts/lcp_subparts.py when: You need to perform LCP (Largest Contentful Paint) subpart decomposition on slow pages.

  10. How DataForSEO integrates with Claude SEO skills

    main

    When the DataForSEO extension is installed, core Claude SEO commands automatically detect its availability and use live data to enhance their output:

    • /seo audit: Uses real SERP, backlink, and on-page data.
    • /seo technical: Uses on-page analysis for real technical data.
    • /seo content: Uses keyword volume, difficulty, and intent data.
    • /seo geo: Uses ChatGPT scraper and LLM mentions for GEO signals.
    • /seo plan: Uses competitor and keyword data for strategy.

  11. Understand Claude SEO data handling and privacy

    main

    Claude SEO is a Claude Code skill that runs locally on your machine. By default, the core skill makes no third-party API calls (though audits will fetch the target URLs you specify) and does not collect, store, or transmit personal data to any vendor.

    Local Execution

    The following processes happen entirely on your local machine:

    • All SEO analysis within your Claude Code session.
    • HTML parsing, content analysis, and report generation.
    • Saving generated reports (PDF, HTML, Excel) to your local filesystem.
    • No telemetry, analytics, or usage tracking is performed.

  12. UCP relationship with existing commerce surfaces

    main

    UCP does not replace existing SEO or commerce tools; it complements them. Understanding these relationships is key to a successful implementation:

    • Google Merchant Center feeds: These are a required upstream dependency. UCP capabilities reference products by the IDs established in your Merchant Center feeds.
    • Product Schema (hasMerchantReturnPolicy, shippingDetails): These are complementary. Schema exposes data at the page layer (for humans/crawlers), while UCP exposes the same data at the API layer (for agents).
    • Google Business Profile (GBP): These are independent. GBP manages store/location data, while UCP manages product/order data.
    • AP2 (Agent Payments Protocol): These work in pairs. UCP handles discovery and checkout structure, while AP2 handles the cryptographic proof of user consent for autonomous purchases.