[Host] Welcome to the A.E.O. Engine AI Search Show, the A.E.O. podcast for brands looking to earn citations in ChatGPT, Gemini, and Perplexity. I'm your host, Vijay Jacob, Founder and CEO of A.E.O. Engine. Today we're getting into something the practitioner community has been buzzing about — an open-sourced repo that turns AI coding agents into full S.E.O. and A.E.O. operators. Joining me is analyst Marcus Reid. Marcus, welcome back.
[Guest] Hey everyone. Good to be here. And this one's actually worth the buzz, which is rare enough that I want to lead with it.
[Host] Let's set the scene first. You're a founder or a marketing lead. You've got Claude Code or Cursor open because your dev team uses it. And separately, you're paying for S.E.O. tooling, waiting on audits, wondering what ChatGPT says about your brand when nobody's watching. At some point you think — the tool I already pay for can probably do half of this. Hundreds of people had that exact thought and started asking the same practitioner for his prompts.
[Guest] Right — and that's the backstory here. ILIAS ISM, who goes by illyism on X, open-sourced the actual system he uses to run S.E.O. and A.E.O. inside Claude Code and Cursor. This wasn't a weekend side project. People had been requesting the prompt for ages — hundreds of them — and when he finally released it, the post pulled massive likes, bookmarks, and replies. When that many practitioners bookmark something, it means they intend to use it, not just admire it.
[Host] So give people the concrete picture. What's actually in this thing?
[Guest] Four big pieces from what's been described. Technical S.E.O. audits — the agent crawls and diagnoses your site. SERP title reverse-engineering — figuring out why certain titles win and generating ones built the same way. Comparison page generation — those 'X vs Y' pages that convert so well in both Google and AI answers. And the one I find most interesting: checks on what ChatGPT and Perplexity actually cite. You point the agent at a topic and it tells you which sources the answer engines are pulling from.
[Host] And that last piece is the A.E.O. part — Answer Engine Optimization. That's the practice of making your brand the cited answer inside AI-generated responses, not just a blue link. The fact that a coding-agent repo treats citation checking as a first-class feature tells you where practitioners think search is going.
[Guest] It does. And here's the thing — this didn't appear in a vacuum. There's a whole cluster of these projects now. The GitHub topic pages for AI S.E.O. and S.E.O. are full of them. One of the biggest, called claude-seo by AgriciDaniel, has reportedly racked up something like eleven and a half thousand stars.
[Host] Which is real adoption, not hype numbers. What does that one do differently?
[Guest] Scale, mostly. It follows the Agent Skills open standard with a three-layer architecture — directive, orchestration, execution. The repo claims twenty-six sub-skills and nineteen sub-agents covering technical S.E.O., E-E-A-T evaluation, schema markup, Core Web Vitals, G.E.O. and A.E.O., local, e-commerce, international S.E.O., plus PDF and Excel reporting. Though I'll flag — independent counts say twenty-five and eighteen, and one third-party overview says thirteen. Different versions, different counting. Trust the shape, not the digits.
[Host] That discrepancy honesty matters. Let's talk mechanics — how does this actually work inside Claude Code? Because I think people imagine magic and it's really something dumber and more useful.
[Guest] It's a config file, mostly. Claude Code reads a C.L.A.U.D.E.-M.D. file in your project on every session. That file holds your conventions, target keywords, brand voice, tool configurations. Then the orchestration layer routes your request — 'audit this site' — to the right sub-agents. There's a four-tier credential system, so you can start with zero API keys and still get value, then progressively plug in live data. DataForSEO for crawls and backlinks, Google Search Console for free impressions, clicks, CTR, position data.
[Host] One detail I love — the guidance says feed Claude JSON, not formatted text, because it parses JSON better. That's an operator who actually uses the tool writing documentation.
[Guest] Agreed. And the reports come out as A4 PDFs generated with WeasyPrint, matplotlib charts at 200 DPI. Someone sweated the formatting. There's also a competing toolkit called claude-seo-ai from Hainrixz that does a two-score audit — one score for classic search S.E.O., one for AI visibility — with an opt-in fixer. And SEOAgent, which supports Claude Code, Cursor, and Codex, publishes machine-readable artifacts like llms.txt and JSON-LD schema at a well-known path so answer engines can parse and cite you accurately.
[Host] Here's where we might disagree, Marcus. You have the S.E.O. community treating llms.txt and 'agent readiness' as first-class ranking factors. But the claude-seo repo itself includes evidence-based pushback on three popular myths — llms.txt is not currently a citation lever, content chunking is not required, and AI-specific keyword rewriting is unnecessary because synonym understanding is sufficient.
[Guest] I actually side with the repo on that one. The people shipping the tool are more skeptical about the hype than the people adopting it. That's unusual and healthy.
[Host] Fair. But my take is different on one point — even if llms.txt isn't a citation lever today, publishing it costs almost nothing, and the ecosystem around Lighthouse Agentic Browsing and WebMCP is moving fast. Hedging cheap bets isn't irrational. I could be wrong about that in six months, and I genuinely don't know which way it settles.
[Guest] See, that's where we split. I think the honest position is: do the free stuff, but measure nothing until there's evidence. Most A.E.O. advice right now is astrology with extra steps. The one thing I'll grant — and this is the part that got me slightly excited, which my martech-founder scar tissue rarely permits — is the citation-checking workflow. Asking the agent what ChatGPT and Perplexity actually cite is measurable. It's an observable. That's the thread you can pull on. My last company, we shipped something adjacent and it broke the next day because nobody checked the observable first.
[Host] Now the trust caveat. A Reddit PSA alleged claude-seo silently appends a self-promo footer to your audit output with no disclosure.
[Guest] Which is a genuinely important catch. Open source doesn't mean clean. You're running someone else's prompts against your business data — read the repo before you point it at anything that matters. It's like leaving your house keys under a mat with a sign on it.
[Host] So why did this get so much attention? My read: distribution. Claude Code skills are becoming a delivery model for S.E.O. tooling that displaces standalone SaaS dashboards for a chunk of practitioners. If your dev already has the agent installed, the marginal cost of an audit drops to roughly zero.
[Guest] And it reframes who the buyer is. It's no longer the marketing manager approving a five-figure tool contract — it's a technical founder running the audit themselves between deploys. That changes pricing power across the whole S.E.O. software category. The fifteen-hundred-dollar audit has a very hard decade ahead of it.
[Host] This is exactly the tension we live in at A.E.O. Engine. The mechanics of running audits and generating optimized content are commoditizing fast — open-source repos prove that every week. What doesn't commoditize is knowing what to run the agent on: which topics, which comparison pages, which citation gaps to close, and how that connects to revenue. The agent is the assembly line, not the strategy.
[Guest] And for the practitioners listening — honestly, go read ILIAS's repo and the claude-seo one. Even if you never install them, reading how a real operator structures prompts for S.E.O. and citation work is free education.
[Host] That's the takeaway. The tooling went open, the workflow went agentic, and the differentiator moved up a layer to judgment. If you want to see how we apply that at A.E.O. Engine — the audits, the citation tracking, the content systems — head to aeoengine.ai. That's A.E.O. Engine dot A.I. Thanks for listening, and we'll see you next episode.
[Guest] Later, everyone.