Router
using-semantic-seo-geek
Routes a broad or multi-part request to the right specialists.
Semantic SEO skills
semantic-seo-geek: an SEO practice for Claude Code and Codex
semantic-seo-geek is an Apache-2.0 licensed plugin that gives Claude Code and Codex a semantic SEO practice instead of a checklist: entity and attribute modeling, topical map architecture, technical SEO review, and evidence-controlled content work, packaged as 11 specialist skills. I wrote it because the methodology I use on client audits needed to exist as software a reader could run, not just a claim on this site.
License: Apache 2.0v1.0.0Language: Python11 specialist skills
The problem it solves
A checklist tells an agent what to check, not what a page is
Most SEO skills for coding agents hand back a checklist: title length, heading count, alt-text presence. semantic-seo-geek decomposes a topic into entities, attributes, and values before any page, title, or schema decision, the entity-first modeling step a checklist skips entirely.
Every material claim the plugin's workflows produce carries one of four evidence labels, observed, sourced, inferred, or unknown, so a recommendation never reads as more certain than the evidence behind it actually is.
What it does
11 specialist skills, one router
A router skill reads a request and calls the smallest useful set of specialists; each specialist below can also be named directly once you know which workflow you need.
technical-seo
Reviews crawling, indexing, rendering, directives, canonicals, sitemaps, redirects, links, and structured-data delivery.
topical-map-architect
Plans site hierarchy, topical coverage, page relationships, URLs, and internal links.
eav-optimizer
Models entities, attributes, values, relationships, and Schema.org mappings the evidence actually supports.
algorithmic-writer
Drafts or revises search-aware copy from an approved brief and source pack.
content-auditor
Finds evidence, usefulness, clarity, and policy problems without automatically rewriting the page.
content-humanizer
Refines approved copy for natural voice and rhythm while preserving claims and qualifiers.
page-production
Assembles a publish-ready page from approved evidence, copy, metadata, links, and media requirements.
title-heading-optimizer
Improves title elements, main titles, heading structure, and title-heading alignment.
visual-semantics
Plans image purpose, alt text, image discovery, licensing details, and visual accessibility.
aiseo-strategist
Assesses source clarity, crawler controls, citations, measurement, and reputation across AI-mediated search.
Four evidence labels on every material claim
Editorial workflows carry these labels into an uppercase claim ledger; an unverified claim becomes UNSOURCED and stays out of publishable copy rather than getting promoted by accident.
- Observed
- Directly verified in a supplied artifact, page, report, dataset, or first-party record.
- Sourced
- Supported by a cited current primary source.
- Inferred
- Reasoned from evidence, with the reasoning stated.
- Unknown
- Not verified, with the missing evidence identified.
Project status
v1.0.0, 1 release, checkable on GitHub
semantic-seo-geek is Apache-2.0 licensed, written in Python, and tagged at v1.0.0 across 1 release on GitHub as of 2026-09-01. It ships as an agent plugin rather than an npm package, so there is no download figure to quote; the release tag and the repository are the checkable record instead, and this section says so rather than borrowing a number from somewhere it does not apply.
License: Apache 2.0Language: Pythonv1.0.01 release
Stats as of 2026-09-01.
Who this is for
SEO practitioners already running Claude Code or Codex
The plugin fits an SEO consultant, an in-house team, or an agency that already works inside Claude Code or the Codex CLI and wants a repeatable practice for auditing, planning, writing, or reviewing search-focused work, instead of prompting the agent from scratch each time.
It adds no crawler, no rank tracker, no API key, and no account. It works from artifacts you supply and the fetching the coding agent can already do, and each workflow says so when a question needs crawl data or measurement it does not have.
How this relates to the practice
The proof the methodology is real enough to ship as software
The entity and attribute modeling behind Technical SEO Audit and AI search visibility work is the same modeling this plugin packages as software, not a private technique described only in a sales conversation.
This site's own content pipeline runs on it. Every page here, including this one, passes through the plugin's content-humanizer scan before it ships, the exact gate this repository's own content pipeline contract names by script path. A methodology that cannot pass its own tool on its own site would not be worth shipping to a client.
Install
Install it in Claude Code or Codex
The commands below are verbatim from the package's own README. Both agents load the same skill bodies; only the manifest differs.
Claude Code
claude plugin marketplace add nipun-arora/semantic-seo-geek
claude plugin install semantic-seo-geek@semantic-seo-geekCodex
codex plugin marketplace add nipun-arora/semantic-seo-geek --ref main
codex plugin add semantic-seo-geek@semantic-seo-geekOpen a new session (or reload the active one) after installing, then describe the artifact and decision you need instead of a special prompt format: "audit this site's crawl and index controls" or "build a topical map for this documentation set from the supplied inventory" both route to the right specialist on their own.
Questions about the plugin, or want the same semantic SEO methodology run on a client engagement?
Work with me