All notable changes to this site are tracked below.
Entries below include only commits on origin/main (already pushed).
Week 15 (From 2026-04-06 to 2026-04-12)
Added
- Experiments index now has live filtering by theme and tool, letting visitors instantly narrow down experiments without a page reload.
Fixed
- XML file URLs are now excluded from the sitemap to improve crawl quality.
Week 12 (From 2026-03-16 to 2026-03-22)
Added
- IndexNow integration: sitemap URLs are automatically submitted to search engines on every successful deploy.
- YouTube shortcode using
youtube-nocookie.comfor privacy-respecting, responsive video embeds on any page. - AI prompt shortcode with multiple visual skins for displaying prompts inline in experiment posts.
- Paginated PPTX presentation-preview shortcode for embedding slide decks as interactive page-flipping previews.
Changed
- SEO indexing narrowed: blog, changelog, and AI landing sections are now excluded from search engine indexing to focus crawl budget on experiments and core pages.
Fixed
- Core Web Vitals CLS score dropped from 1.055 to ~0 by switching to a standard blocking stylesheet instead of async CSS loading.
- Page titles shortened to stay within Bing and Google’s recommended 65-character limit.
- Canonical hygiene: noindex pages are now excluded from the sitemap, and paginated URLs enforce a trailing slash via redirect.
Week 11 (From 2026-03-09 to 2026-03-15)
Added
- Published experiments: Do Claude’s Usage Limits Push Users to Codex?, Can AI Help Me Get Backlinks Without Turning Into a Spammer?, Should I Add a Newsletter If It Costs Me Website Traffic?, Can I Finally Scrape My X Bookmarks into New AI Experiments?, What Do Claude’s New Diagrams Reveal About My Experiments?, Do We Still Need PRD Documents When AI Can Code?, Can Henk Read My Archive and Write in My Voice?.
Changed
- Homepage intro copy now adds a final line framing the experiments as self-discovery before the primary CTAs.
- About page copy now explains the personal motive behind the daily experiments and the internal obstacles they reveal.
Fixed
- SVG resources in content image rendering now bypass Hugo raster-only dimension access, and the Experiment 25 site-structure diagram was reduced to static SVG markup so it loads without inline behavior.
Week 10 (From 2026-03-02 to 2026-03-08)
Added
- Experiment 14 published: “Can I run a decent local AI model on my iPhone 17 Pro?”.
- Published experiments: Can I Run a Decent Local AI Model on My iPhone 17 Pro?, How Can I Optimize My Website for LLMs and Generative AI?, Scrape X Bookmarks for New AI Experiment Ideas, Optimize LinkedIn Posting Strategy via Analytics, Could I Create a GitHub-Style Activity View as a Hugo Module?, Building My Own CRM: Can the Vision Survive?, Can I contribute a Hugo component to open source?.
llms.txtguidance file at the site root for LLM crawlers.- Footer/profile links to X and LinkedIn.
Changed
- Site-wide content images now run through Hugo image processing (responsive
srcset+ compressed WebP variants) for markdown images and gallery/inline image shortcodes. - Experiment 13 now supports compact prompt previews from local
prompt*.mdfiles via a newprompt-previewshortcode, with consistent wrapped rendering for long prompt text. - Site-wide processed WebP image quality was retuned to
q94for sharper text and UI screenshots. inline-imagesshortcode now supports optionalqualityandwidthparameters, with width-constrained image blocks aligned to the left edge of the text column.- Plan documents are now colocated inside their experiment folders (page bundles), and the separate
/planssection/templates were removed. - Added schema.org JSON-LD across Hugo-rendered pages (
WebSite,Person,WebPage,CollectionPage,ItemList,BlogPosting) plusService/Offer/FAQPagemarkup on the OpenClaw setup landing page. - Experiment pages now use
Articleschema, with strengthened publisher/organization/social relationships across structured data. - SEO/discovery metadata was expanded across pages: canonical usage, sitemap reference in robots output, breadcrumbs markup, and fuller social/author/article meta tags.
- Performance pass shipped: non-blocking main stylesheet loading, lazy/deferred comments and hotkeys assets, and deferred interaction initialization on the OpenClaw service page.
- Email exposure was reduced by obfuscating contact links at render/runtime and removing direct email fields from schema output.
- Navigation/mobile UX updates:
Homehidden on mobile,AI Experimentsshortened toExperimentson mobile, and dark-mode toggle alignment tuned with nav buttons.
Fixed
plan-previewshortcode heading hierarchy now avoids renderingh1.- Mobile nav regression where
Homeremained hidden at larger breakpoints.
Week 9 (From 2026-02-23 to 2026-03-01)
Added
- Site icon set and explicit
410 Gonehandling for invalidup-nextpaths. - Published experiments: Finding the Right Local AI Model to Run on My Hardware, Troostwijk Auction Search Sucks, Let’s Fix That, Can I Build a Local Transcript and Summary Menu Bar macOS App, Does HeyGen Help Me Create a Video Faster?, How Does Nano Banana 2 Model Do Text in Images?.
Changed
- Experiment metadata/title updates and day/date metadata rows on overview cards.
Fixed
- Memoji avatar crop and icon asset regeneration.
Week 8 (From 2026-02-16 to 2026-02-22)
Added
- Custom comment system with reactions, spam controls, moderation workflow, and admin UI.
- Moderation enhancements: ntfy push alerts plus manual test action in dashboard.
- Content presentation features: reusable gallery, Experiment 2 page bundle migration, and Experiment 1 plan preview.
- Rich post content support: X embed shortcode with widget loader, inline image shortcode, and quote styles.
- Published experiments: Experiment 1: Let Codex Build the Lab, Use ChatGPT for Pruning, Add Custom Discussion Tooling to hugomelis.nl, Agents Are Taking Over the World, Meet Henk, My Bot Can Order Groceries Now, We Need That Right?, Moltbook: A Shared Notebook Between Bots, Running a Local Model to Allow My Thoughts to Flow Freely, Can Henk (OpenClaw) Fix Home Assistant Problems?.
Changed
- Homepage/navigation copy and CTA polish, plus iterative experiment content/layout updates.
- Production delivery hardening:
wwwto apex redirect, production base URL updates, and analytics migration from Plausible to proxied Umami.
Fixed
- Comment PoW normalization mismatch and experiment draft/date cleanup.
Week 7 (From 2026-02-09 to 2026-02-15)
Added
- Initial Hugo v1 foundation with core pages/templates, taxonomy/pagination, visual styling system, and theme toggle.
- Deployment setup and operational baseline: Netlify configuration, analytics wiring, documentation, and starter assets.
- AI Experiments section launch with homepage hotkeys, first experiment entries, and day metadata on experiment cards.
Changed
- Site-wide layout/branding/content refinements across home, blog, about, and now pages.
- Homepage positioning was updated to the daily AI experiment direction.
Fixed
- Container width consistency, blog header alignment, markdown list rendering, and experiment day-pill wrapping/alignment.