Agentic SEO Skill

git clone https://github.com/Bhanunamikaze/Agentic-SEO-Skill && cd Agentic-SEO-Skill && bash install.sh

What it does

Runs structured SEO and GEO audits the way a senior consultant would, then backs every finding with evidence. It ships 16 sub-skills (full audit, single page, technical, schema, sitemap, GEO, AEO, hreflang, GitHub repo SEO, and more), 10 specialist agents, and 89 scripts that collect the raw data so the verdicts are not guesswork. A full audit writes a detailed report plus a prioritized action plan.

VeryCreatives' take

Most SEO skills just lecture the model about best practice. This one gathers proof first. Behind the 16 sub-skills sit 89 scripts that pull real evidence: Core Web Vitals and PageSpeed data, crawl results, robots and sitemap checks, indexability verdicts, schema detection. The findings are backed by data instead of the model's confident guesses, every fix comes back prioritized and confidence-labeled, and a full audit writes both a detailed report and an action plan you can actually work through.

The deterministic trigger mapping is the quiet reason it holds up. "perform seo analysis on <url>" always runs the same path, so you get a repeatable audit rather than a different answer each run. That reliability is the difference between a skill you can put in front of a client and a one-off party trick.

It is also unusually current. Alongside classic technical SEO it covers GEO and AEO, the answer-engine and AI-search optimization most SEO tooling has not caught up to yet, plus hreflang, E-E-A-T, and even GitHub repository SEO. For a category moving as fast as AI search, that breadth matters.

The honest limit: it is heavy, and it is an analysis skill, not a do-it-for-me one. The evidence layer is 89 Python scripts and several need API keys, so budget setup time, and without that layer you lose the main advantage. It tells you what to fix and why; you or your team still do the fixing.

Bottom line: if you do SEO or GEO seriously, this is a top pick, and the evidence layer plus the GEO coverage put it well ahead of a generic SEO prompt. If you want an outside set of eyes on what it surfaces, that is what the VeryCreatives SEO team does.

How VeryCreatives uses it

This skill is part of our daily SEO workflow, not a tool we tried once. We run it in Claude Code as part of the weekly SEO work on our own site, and we have used it to move pages up the SERPs.

We reach for it across the full range of our SEO work: client audits, GEO and AEO checks, technical SEO, content and E-E-A-T review, and content gap analysis. It is also one of the better ways we have found to turn an audit into a clean, client-facing report.

The part worth calling out: we do not run it standalone. We built our own MCP server that wraps the Google Search Console API along with third-party keyword research, keyword clustering, and search-trend analysis APIs, and we run this skill on top of that. The skill brings the structured, evidence-backed audit logic; our MCP brings the live ranking, keyword, and trend data. Together they are far more powerful than either on its own, and that combination is how we get from a raw audit to decisions we can act on.

If you already have your own SEO data sources, that is how we would point you: keep the skill for the audit discipline and feed it your real data, rather than treating it as a closed box.

When to use it

• You want a repeatable, evidence-backed SEO or GEO audit, not a generic checklist.
• You work across agents (it ships native install formats for several IDEs).
• You care about AEO and GEO, not only classic technical SEO.

When not to

• You cannot run the bundled Python scripts or supply the API keys they need (see setup). Without the scripts you lose the evidence layer that makes this skill worth choosing.
• You only need a one-off title-tag check; this is heavier than that.

Setup note

The 89 scripts are Python and several expect API keys (set via a .env file, see .env.example in the repo). This requirement is the same on every agent, so it is not an agent-specific caveat, but budget a few minutes to wire it up before the first audit.