SEO Landing gives coding agents a technical SEO checklist as a skill
SEO Landing is an agent skill that gives Claude Code, Cursor, Codex and six other coding agents a technical SEO checklist targeting a 100/100 PageSpeed score, a 15 KB JavaScript budget, and Core Web Vitals inside the green thresholds, at 136 stars in its first week on GitHub.
Why it mattersA landing page generated by a coding agent is often published with the performance and metadata problems SEO consultants spend hours fixing later, and packaging the checklist as an agent skill fixes those problems in the same process that created them.
aleksandr-alhoff/seo-landing was created on 28 August 2026 and has reached 136 stars and 11 forks in its first week on GitHub. It is an agent skill, MIT-licensed, that turns an AI coding agent into a technical SEO specialist for landing pages. The author, Aleksandr Alhov, describes himself as an SEO consultant with fourteen years of experience across SEO, content marketing, and GEO and AEO work.
What the skill enforces
The skill lists explicit performance targets the agent has to hit before a page is published. The stated PageSpeed goal is 100 out of 100, and the Core Web Vitals thresholds are LCP under 2.5 seconds, INP under 100 milliseconds, and CLS under 0.1. The JavaScript budget for the first load is capped at 15 kilobytes, and the first load is required to make zero third-party requests. Structured data covers five Schema.org types: WebSite, Organization, BreadcrumbList, FAQPage, and VideoObject.
The rest of the skill is the technical discipline behind those numbers. Images are converted to AVIF with responsive variants. Fonts load asynchronously with a font-display strategy that avoids layout shift. Critical CSS is inlined and the rest is deferred. Semantic HTML and WCAG 2.1 AA accessibility are treated as non-optional. The stated purpose of packaging all of this as a skill is that a coding agent without guidance tends to produce pages with heavy JavaScript, unoptimised images, weak metadata, and missing structured data. That list matches the defects an SEO auditor spends most of an engagement fixing.
Which agents it runs in
The README lists nine coding agents the skill is compatible with: VS Code Copilot, Claude Code, OpenAI Codex, Google Antigravity, Cursor, Gemini CLI, OpenClaw, Hermes, and one described only as a general skill runner. That is a wider range than most agent skills claim, and it reflects a design that reads as a plain checklist and prompt structure rather than tool calls specific to one runtime.
Landing pages have become the class of page that an AI coding agent is most likely to generate end to end, because they are visually distinctive, short, and change often. That is also the class of page where the gap between "the browser renders it" and "search engines and answer engines can use it" is easiest to miss. A page that looks fine can still take four seconds to become interactive on a real phone, load 400 kilobytes of JavaScript that does nothing in the first screen, and carry no structured data at all, and none of that shows up in a screenshot.
The value of the SEO Landing project is where the rules are placed: inside the coding agent that writes the page. The specific list of thresholds is one any competent audit would produce. A skill file the agent reads before it starts is a cheap way to remind the model of the constraints while it writes the page, rather than fixing the output afterwards. Whether the resulting pages actually meet the 100 out of 100 target the README claims is worth measuring on real work rather than trusting.
Source
This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.
Get AI News in your inbox
New developer tools, model and agent releases, and how teams are actually using them to release software. Short, and only when there is something worth reading.