Built SEO agents for Claude and Cursor to fix my own mess. What am I missing?
Hey Product Hunt 👋
Maker of SEO Skills AI here. Quick backstory: I didn't plan this as a product. I built it because my own SEO workflow was a disaster.
I run a few niche sites, and my setup was painful. Expensive keyword tools, a separate audit tool, spreadsheets everywhere, hours of copy-pasting between tabs. Meanwhile Claude Code and Cursor were open on my screen all day anyway. One night it clicked: why am I paying for five tools when the AI in my terminal could do most of this?
So I wrote my first SEO skill for Claude Code. It found low-competition keywords in under a minute. Then I made one for content optimization. Then audits. Then meta tags. A few weeks later my whole SEO workflow ran inside my coding assistant, and I cancelled the subscriptions.
That's seoskillsai.com. Plug-and-play SEO agents for Claude, Codex, Cursor and Antigravity. Keyword research, content briefs, on-page optimization, technical audits, internal linking.
It fixed my problem. But your workflow isn't mine, so tell me what to build next:
🔧 Which assistant should I support next? Windsurf, Copilot, Aider, Cline, Zed?
🔌 MCP server with live tool access, or keep skills as simple drop-in files?
📊 Data integrations (Search Console, Ahrefs, Semrush APIs) or keep it API-free and free?
🌐 Should skills publish straight to WordPress and Webflow, or stick to analysis and generation?
🗺️ What's missing completely? Local SEO, schema, rank tracking, multi-language? What's the one SEO task you still hate doing by hand?
I'm here all day. Every request in this thread goes on the public roadmap, and the most requested one gets built first. I'll post updates in the comments.
Github: https://github.com/seoskillsai/seo-skills-ai
Thanks for reading 🙏
Replies
sounds cool, going to test it later today
Founder here, I run a done-for-you Amazon affiliate site service, so this is the publisher side of your question.
Taking your last question straight: the SEO job I still hate doing by hand is proving that a ranking turned into anything. Ranking is cheap to measure and everything downstream of it is manual. My own pricing page sits at position 3 for its main term, and nobody searches that term, so position 3 buys nothing. On the affiliate side, one market once sent 1375 outbound clicks to Amazon and produced zero orders. Every dashboard reported both as normal. None of them told me whether I had a keyword problem, an intent problem or a page problem, and that triage is where the hours actually go.
The bigger gap I would point your agents at: on one of my sites, ChatGPT has overtaken Google as the number one referring source. That traffic does not appear in Search Console at all. An agent wired only into GSC is optimizing the smaller half of discovery and will confidently tell you to fix pages that are already being cited elsewhere. Pulling referrer data alongside GSC would have changed what I worked on for the last two months.
One more thing specific to affiliate and ecommerce pages that generic SEO tooling misses entirely: freshness is not a content quality nicety, it is the ranking mechanism. A comparison page with a stale price or a discontinued product keeps its position for a while and converts at zero, so the traffic graph tells you nothing is wrong until the position is gone too. If your agents can flag pages whose underlying data has drifted since publication, that is worth more than another title tag suggestion. Question back at you: when your agent finds a page ranking well, does it have any way to judge whether that ranking is worth defending?
@serdarships Thank you for your input. I will test this approach on one of my projects and try to implement it