Just dropped a demo of our new Product Page Kit. A ChatGPT voice agent audits a real product page live in Chrome, points at every fix with the cursor, and builds a PDF action plan. Watch it here: https://www.youtube.com/watch?v=A6dL7PpoeqI The kit is Pro-first for now, same as always. It will reach you here at the right time, bit by bit, like every skill before it. Pro members have it today.🆕

Fluent is not the same as true. That is the whole problem with AI-assisted content, and it is why "does this sound like AI" is the wrong question. I gave Gemini one prompt: "write me a blog about medicine." It came back polished. Then I ran it through anti-slop, the skill I built, and it returned 14 defects with a verbatim quote and a mechanical test behind every one. You can run three of those tests by hand right now, on anything you published this week. 1. The deletion test. Cut a sentence. Say out loud what was lost. If the answer is "nothing", it was padding. 2. The inversion test. Negate the claim and write the negation out in full. "This plays a crucial role in the industry" becomes "this plays no role in the industry." If nobody would ever assert the negation, the original said nothing. 3. The attribution test. Every "studies show", "experts say", "industry reports" gets resolved to a named source, and then you check that the source says that specific thing. There is no third outcome. It resolves or it does not. Most AI-assisted marketing copy fails all three, and so does a lot of human-written copy. That is the point. The tool never tells you who wrote something. It tells you which claim does not hold up, which is the only thing you can actually fix. The full walkthrough, including the install: https://www.youtube.com/watch?v=jxTGZEcvtsU Repo, free and open source: https://github.com/AgriciDaniel/anti-slop Run test 2 on your last landing page headline and drop the negation in the comments. If nobody would argue with the opposite, the headline is empty.

The workflow JSON is attached. Here's what it does and what you'll need. Example of output One form input: website URL, country, language. From that it runs six phases: 1. Keywords the site already ranks for (volume, intent, difficulty, competition, CPC, bids, trends) 2. Keywords the site could rank for 3. Google Lighthouse, full audit 4. Sitemap map + a scrape loop over every page (screenshot, HTML, markdown) 5. On-page SEO audit 6. Domain signals: backlinks, domain rank, technologies, relevant pages Then it renders one canonical report object into a PDF, an HTML report, a Markdown file and an XLSX keyword sheet. What you need before importing: - n8n (self-hosted or cloud) - DataForSEO API key - Firecrawl API key - Gotenberg running for the PDF render (this one trips people up - the PDF node points at `http://gotenberg:3000`, so it needs to be reachable from your n8n container) Full video walkthrough: https://www.youtube.com/watch?v=O53pNhQopPI Import it, run it on your own site, and post your report's overall health score below. If a node errors, paste the error and which node it was and I'll help you fix it.

The public repo's been sitting on an older version for 2 months. Today I pushed everything I've built since, live: final tests, the changelog, version tagging, and the actual GitHub release. No edits, no dry run. If you write content, run a blog, do SEO, or just want to watch a real open source release happen from start to finish, this is for you. I'll be answering questions live as it happens, so bring whatever you're curious about, whether it's how the skill works, how to install it once it's public, or how the release process itself goes. Watch here: https://www.youtube.com/live/Eo-mv6JE9MQ

We shipped the update live: moving the release-ready improvements from our private development repository into the public open-source edition, reviewing the changes, running the checks, and publishing the release. What changed: ✅ Fresher Google Search and AI Search guidance ✅ More accurate commands, documentation, and recommendations ✅ Stronger consistency, portability, security, and privacy checks ✅ Cleaner prompts and fewer outdated false positives ✅ No breaking changes — the existing workflow remains the same Final validation included 328 passing tests, clean consistency and portability checks, and successful credential, PII, and private-reference scans. Useful links: 🎥 Watch the livestream replay: https://www.youtube.com/live/aHJwrswC3ko ⭐ Claude SEO on GitHub: https://github.com/AgriciDaniel/claude-seo 📦 Claude SEO v2.2.3 release: https://github.com/AgriciDaniel/claude-seo/releases/tag/v2.2.3 📎 I’ve also attached the presentation covering the latest Google updates, what they mean for Claude SEO, and the release workflow we used during the livestream. If you use Claude SEO, grab v2.2.3 and let me know what you build with it! 🙌