What does this skill do, and when should you use it?
This skill clusters keywords by shared URLs in Google’s top organic results rather than by textual similarity. It expands a seed keyword, classifies search intent, and designs pillar-and-spoke content architecture. It also produces an internal-link matrix and an interactive cluster map. With claude-blog installed it can execute content creation; otherwise it generates detailed content briefs.
Reads a seed keyword or an existing SEO strategy file; expands the seed into 30–50 variants using WebSearch, related searches, PAA questions, and intent modifiers; compares candidate keywords by shared URLs among the top ten organic results; removes navigational terms and classifies informational, commercial, and transactional intent; selects pillar keywords, spoke groups, templates, and word-count targets; creates bidirectional internal-link plans; and writes cluster-plan.json, cluster-plan.md, cluster-map.html, plus cluster-briefs/ when claude-blog is unavailable. An optional DataForSEO MCP integration can provide SERP data.
- An SEO consultant is planning a new site around one core topic and needs pillar and supporting-page recommendations.
- A content team has a keyword list and wants to decide which terms should target the same page based on actual SERP overlap.
- A site owner needs to identify cannibalization risks, orphan pages, and missing internal links in a planned content structure.
- A team with claude-blog installed wants to create the pillar page and supporting posts in priority order.
- A team without a content-generation skill needs structured briefs containing titles, metadata, outlines, and linking requirements.
How do you install this skill?
- Core clustering depends on Google WebSearch or DataForSEO; reachability and suitability for mainland-China networks, Chinese keywords, and localized SERPs are unproven.
- Execution writes to the current directory and may create content; staged confirmation, isolated output paths, backup, and rollback are not clearly specified.
- The metadata version 2.2.0 conflicts with the document heading version 1.9.0; referenced scripts, templates, and reference files were not individually verified in the supplied target-path evidence.
- SERP results, search geography, and rankings can change clustering conclusions; manually review intent classification, the overlap matrix, internal-link plan, and generated content.
- Shell / CLI
- Network access
- Local filesystem
- MCP Server
Python 3.10+
The skill is bundled in AgriciDaniel/claude-seo at skills/seo-cluster/SKILL.md. For Claude Code plugin installation, run /plugin marketplace add AgriciDaniel/claude-seo, then /plugin install claude-seo@agricidaniel-claude-seo. Alternatively, run git clone --depth 1 https://github.com/AgriciDaniel/claude-seo.git followed by bash claude-seo/install.sh.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/AgriciDaniel/claude-seo.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/seo-cluster" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Run /seo cluster plan <seed-keyword> for expansion, clustering, architecture, and visualization. Use /seo cluster plan --from strategy to import existing /seo plan output. Use /seo cluster execute to execute the plan or generate briefs, and /seo cluster map to regenerate the visualization from cluster-plan.json. When using a URL as the seed, the workflow must be able to retrieve SERP data.
What are this skill's strengths and limitations?
- Uses shared Google organic-result URLs as the clustering signal instead of relying only on phrase similarity.
- Produces actionable pillar, spoke, template, word-count, and internal-link specifications.
- Includes cannibalization, orphan-page, and link-coverage checks.
- Exports JSON, Markdown, HTML, and detailed briefs when claude-blog is absent.
- Requires WebSearch or optionally DataForSEO for SERP data; if both fail, it falls back to intent-only clustering with a warning.
- The instructions use Claude Code /seo command behavior, so another client would require adaptation.
- Automatic content creation requires claude-blog; otherwise the output is a set of content briefs.
- The supplied material does not provide independent test results or cross-platform validation for this individual skill.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| SERP Semantic Topic Clustering this page | 49 · Use with care | ★ 19k | 23d ago | MIT |
| Blog Cluster Engine | 53 · Use with care | ★ 2.4k | 14d ago | MIT |
| SEO Content Brief Generator | 52 · Use with care | ★ 19k | 23d ago | MIT |
| Blog Outline Planner | 39 · Not recommended | ★ 2.4k | 14d ago | MIT |
| DataForSEO Live SEO Intelligence | 47 · Use with care | ★ 19k | 23d ago | MIT |
How did FollowSkills review this skill?
The skill states that it does not store or transmit credentials and describes URL safety, DataForSEO cost checks, and fallback behavior. However, it invokes external search, reads local strategy files, writes to the current directory, and may create content through claude-blog without clear data-flow disclosure, least-privilege scoping, pre-execution confirmation, or complete rollback, so 11 points are deducted.
The workflow, thresholds, output files, and error table are fairly detailed, with repository-level CI and test evidence available. However, SERP providers are variable, availability of referenced scripts and files is not verified within the target skill, retry behavior is vague, and the metadata version 2.2.0 conflicts with the document version 1.9.0; points are deducted within the static-review cap.
Audience, trigger phrases, argument hint, and primary scenarios are clear, and the skill accepts a keyword or URL plus strategy import. Its core capability nevertheless depends on Google WebSearch or DataForSEO, with no evidence for Chinese-language, localized SERP, or mainland-China network fit and limited non-fit boundaries, so 7 points are deducted.
The documentation has good layering and includes a quick reference, workflows, architecture, output formats, error handling, license, author, and version metadata. It lacks clear target-skill installation/dependency notes, FAQs, changelog, maintenance ownership, and update path; the document and metadata versions also conflict, so 5 points are deducted.
It specifies directly usable workflows and formats for SERP-overlap clustering, hub-and-spoke architecture, internal-link matrices, and visualization, offering clear value over manual planning. Results still depend on live SERPs, keyword volume, and search geography; generated content requires human review, and no target-skill representative output is verified in the supplied files, so 9 points are deducted under the static cap.
The methodology defines overlap computation, thresholds, matrix format, and source-priority guidance. Repository CI and tests provide limited corroboration, but there is no committed execution evidence covering this skill's key paths, and SERP conclusions cannot be confirmed from static text alone, so 6 points are deducted.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →