This website is built with Eleventy (11ty) using Nunjucks templates. All pages are configuration-driven — there are no individual .njk files per page. Instead, a single pagination template (src/pages.njk) generates every page from JSON config files.

1. src/content/site.json defines the navigation menu, footer, social links, and global variables.
2. src/_data/pages.js reads the menu structure and resolves a config.json from each content directory.
3. src/pages.njk uses Eleventy pagination to generate one HTML page per entry.
4. Each config.json contains a blocks array. Each block specifies a templateName and content:

   {
     "templateName": "imageAndTextBlock",
     "content": { "title": "...", "img": "...", "align": "right" }
   }

5. The renderBlock shortcode renders the matching Nunjucks template from src/_includes/blocks/.

src/
├── pages.njk              # Single pagination template (generates all pages)
├── _data/
│   ├── pages.js           # Builds page list from site.json menu
│   └── site.js            # Exposes site.json globally
├── _includes/
│   ├── base.njk           # HTML scaffold (header, nav, footer)
│   └── blocks/            # ~21 reusable block templates
│       ├── carouselBlock.njk
│       ├── imageAndTextBlock.njk
│       ├── galleryBlock.njk
│       ├── textBlock.njk
│       └── ...
├── content/
│   ├── site.json          # Menu, footer, socials, variables
│   ├── config.json        # Homepage blocks
│   └── [section]/
│       ├── config.json    # Page blocks
│       └── assets/        # Page-specific images/media
├── assets/                # Global assets (fonts, CSS, JS, images)
└── build/                 # Build-time helpers
static/                    # Files copied to build root (robots.txt, CNAME, etc.)
build/                     # Output directory (generated)

Assets are served from three sources:

Per-page assets are discovered automatically by parsing the menu structure in site.json.

npm install

Starts a local dev server at http://localhost:8080/ with live reload:

npm start

npm run build

Output is written to the build/ directory.

1. Create a directory under src/content/ with a config.json containing a blocks array.
2. Add a menu entry in src/content/site.json with a root matching the directory name.
3. The page is generated automatically on the next build.

To edit an existing page, modify its config.json — add, remove, or reorder blocks. Available block types are the .njk files in src/_includes/blocks/.

• {{ key | var }} — Resolves a value from site.json's variables object.
• {{ obj | json }} — JSON-stringifies a value.

This repository uses GitHub Copilot skills (.github/skills/) and helper scripts to automate content updates across the site. The blog post at .github/blog_post/BlogPost.md is the single source of truth for all feature content — every skill reads from it. See .github/INSTRUCTIONS.md for detailed workflow documentation.

Before running any update, make sure the blog post is in place:

1. Save the release blog post markdown to .github/blog_post/BlogPost.md.
2. For homepage updates, save feature images to .github/raw_images/ and run npm run process-images.

Open GitHub Copilot Chat and type a natural language prompt. You don't need to reference skill files — Copilot matches your request to the right skill automatically. Examples:

• "Update the specifications page with the new features from the blog post"
• "Add the new demos to the feature demos page"
• "Add alt tags to the main page"
• "Update alt text on the community page"

Each skill will read the blog post, compare against existing content, and present a plan before making changes. You confirm before anything is written.

The homepage skill handles the most complex update — converting blog post features into visual content blocks with images, background colors, descriptions, and links.

Process:

1. Place feature images in .github/raw_images/ (named to match the blog post image references).
2. Run npm run process-images to resize, optimize, and sample background colors. Colors are written to .github/image-colors.json.
3. Ask Copilot: "Update the homepage with new features from the blog post."
4. The skill creates imageAndTextBlock entries with alternating alignment, sampled background colors, condensed descriptions, and doc links.
5. Run npm start to preview at http://localhost:8080.

To update a single image later:

1. Place the updated image in .github/raw_images/ with an _updated suffix (e.g., particleFMAT_updated.png).
2. Run npm run update-image -- particleFMAT.
3. Update the background hex color in src/content/config.json with the value printed by the script.

Ask Copilot: "Update the specifications page from the blog post."

The skill extracts feature names and documentation links, then places each feature in the best-matching section (Main Features, Shaders/Rendering, Geometry, Optimizations, Special FX, Cameras, or Exporters and Tooling). It flags potential duplicates with existing entries and asks how to handle them. After adding all features, it checks whether the left and right columns are balanced and suggests moving whole sections if needed.

Automated screenshot capture of WebGL Playground demos is extremely resource-intensive (headless Chromium rendering full 3D scenes). Instead, the feature demos skill uses a manual screenshot workflow:

Pass 1 — Create entries with placeholders:

1. Ask Copilot: "Add blog post demos to feature demos."
2. The skill identifies non-Playground demo links (e.g., the Babylon.js Editor or Viewer Configurator) and asks whether to skip them.
3. It creates all config entries with placeholder image paths and temporary alt text, then outputs a table of camelCase filenames you need to provide (e.g., clusteredLighting.jpg, frameGraph.jpg, sdfText.jpg).

Between passes — Capture screenshots manually:

4. Visit each demo link from the blog post and take a screenshot.
5. Save each screenshot to src/content/featureDemos/assets/img/ using the exact filename from the table.

Pass 2 — Generate alt text:

6. Ask Copilot: "Add alt tags to the feature demos page."
7. The alt text skill views each image and writes descriptive alt text based on what the screenshot actually shows.

The alt text skill works on any page — not just the homepage. Ask Copilot to add or fix alt tags on any page, and it will find images with missing, empty, or placeholder alt text, view each image file, and write descriptive text.

Examples:

• "Add alt tags to the main page"
• "Fix alt tags on the feature demos page"
• "Update alt text on the community page"

.github/
├── INSTRUCTIONS.md          # Full workflow documentation
├── image-colors.json        # Generated background colors
├── blog_post/
│   └── BlogPost.md          # Blog post content (ground truth)
├── raw_images/              # Source images for homepage
├── scripts/
│   ├── process-images.js    # Batch image processor
│   └── update-image.js      # Single image updater
└── skills/
    ├── homepage-update.prompt.md       # Homepage feature blocks
    ├── blog-text-conversion.prompt.md  # Text conversion rules
    ├── specifications-update.prompt.md # Specifications page entries
    ├── feature-demos-update.prompt.md  # Feature demos gallery
    └── homepage-alt-text.prompt.md     # Image alt text (any page)

this property available as part of page level Json property

    "title": "BabylonJS - 3D engine based on WebGL/Web Audio and JavaScript"

meta tags can be unique for each page. This property available as part of page level Json property and contains list of tags

    "metaTags": [
    {
      "name": "description",
      "content": "BabylonJS - 3D engine based on WebGL/Web Audio and JavaScript"
    },
    {
      "property": "og:title", //optional property for open graph
      "content": "BabylonJS - 3D engine based on WebGL/Web Audio and JavaScript"
    }
  ]