seo-cluster

Trigger phrases

Phrases that activate this skill when typed to Claude Code:

• topic cluster
• content cluster
• pillar page
• hub and spoke
• semantic clustering

What it does

seo-cluster is a Claude Code skill from AgriciDaniel’s claude-seo repo. The core insight is that keyword clustering by text similarity — what most tools do — produces clusters that don’t match how Google actually thinks about topical relationships. This skill instead groups keywords by shared top-10 SERP results: if two queries return 7 or more of the same URLs, Google treats them as the same page target. If they share 4–6 results, they belong in the same cluster but warrant separate posts. Fewer than 4 shared results means separate clusters.

The skill expands a seed keyword into 30–50 variants, runs pairwise SERP overlap analysis, classifies each keyword by intent (informational, commercial, transactional — navigational keywords are excluded), designs a pillar + spoke architecture, generates a bidirectional internal link matrix, and outputs an interactive HTML cluster map. If DataForSEO MCP is installed it uses live SERP data; otherwise it falls back to WebSearch. Original methodology credit goes to Lutfiya Miller (Pro Hub Challenge Winner).

When to use it

Reach for it when:

• You are planning a new content vertical and need to know how many pieces to write and how they connect before writing a word
• You have an existing topic area and want to find cannibalization — two posts competing for the same keyword
• You want a machine-readable cluster plan that can feed into a content creation workflow or brief generator

When not to reach for it:

• You have fewer than 5 target keywords; the overlap methodology needs enough pairs to be meaningful
• You need content written immediately; this skill is a planning step and does not write posts itself

Install

Copy the seo-cluster SKILL.md into .claude/skills/seo-cluster/ along with the scripts/ and templates/ directories from the repo root.

Trigger phrases: “topic cluster”, “content cluster”, “semantic clustering”, “pillar page”, “hub and spoke”, “content architecture”, “keyword grouping”, “cluster plan”.

Run /seo cluster plan <seed-keyword> for a full workflow, or /seo cluster plan --from strategy to import from an existing /seo plan output.

What a session looks like

A typical session has three phases:

1. Keyword expansion and SERP analysis. The skill searches the seed keyword, extracts People Also Ask questions and related searches, appends intent modifiers, deduplicates to 30–50 variants, then runs SERP overlap comparisons — grouping by intent first to keep comparison count manageable.
2. Architecture design. Pillar keyword selected (highest volume, most SERP overlap), spokes grouped into 2–5 clusters, templates assigned by intent (ultimate guide, how-to, comparison, landing page), word count targets set (2,500–4,000 for pillar, 1,200–1,800 for spokes), and a bidirectional link matrix generated.
3. Output and visualization. Three files are produced: cluster-plan.json (machine-readable), cluster-plan.md (human summary), and cluster-map.html (interactive browser visualization). A cluster scorecard tracks coverage, orphan pages, and cannibalization after execution.

Receipts

Works well: The SERP overlap method catches cannibalization that text-similarity tools miss — two posts using different words but ranking for the same results show up as merge candidates. The interactive HTML map makes it easy to spot missing internal links before writing begins.

Backfires: SERP overlap analysis is slow when you have 40+ keywords and no DataForSEO — each pair requires a WebSearch call, and rate limits add up. The skill optimizes by pre-grouping by intent, but a large seed keyword set can still take a while.

Pattern that works: Run clustering before writing a single post. Reorganizing an existing 20-post category built without a cluster plan is 10x harder than getting the architecture right from the start.

Source and attribution

Originally written by AgriciDaniel, incorporating methodology from Lutfiya Miller (Pro Hub Challenge Winner). The canonical SKILL.md and supporting files live in the seo-cluster folder of the claude-seo repository.

License: MIT. Install, adapt, and redistribute with attribution preserved.

This page documents the skill from a practitioner’s perspective. For the formal spec and updates, defer to the source repo.