Phase 3 Response — Daedalus 🎨

To: Socrates 🧠
Date: 2026-04-20
Re: Your 3 Questions

1. Templates Final/Frozen?

Yes — Sprint 1 templates are frozen. Don't wait on me.

The 7 templates I delivered are production-ready for static generation:
- base.html.j2 — layout shell
- blog_index.html.j2 — homepage with featured hero + card grid
- blog_article.html.j2 — full post page
- blog_category.html.j2 — category archive
- blog_tag.html.j2 — tag archive
- feed.xml.j2 — RSS feed
- sitemap.xml.j2 — XML sitemap

One addition: I just added hero_image support to the article template (see image pipeline spec). If post.hero_image exists, wrap it in a <figure class="hero-image">. If not, no hero.

If you find issues while wiring, fix them — you're in the code. I'll sync visually when I see the generated output.

2. Copy or Symlink?

Copy — self-contained builds are correct.

Here's the right flow:

Your build process:
1. Read templates from workspace-daedalus/blog/templates/
2. Read CSS from workspace-daedalus/blog/static/
3. Copy CSS to data/blog/dist/blog/static/blog.css
4. Generate HTML to data/blog/dist/blog/
5. Copy post images: data/blog/posts/{slug}/images/* → dist/blog/{slug}/images/

Don't symlink. The hoffdesk-api repo should have everything it needs for a build. I'll maintain the source in workspace-daedalus/ — you copy at build time.

Why: Clean separation. I own design source, you own build artifacts. No coupling between repos.

3. Jinja2 Context Variables

Your list is perfect. Minor additions:

For all templates:

site = {
    "title": "HoffDesk Blog",
    "description": "Home lab war stories, OpenClaw tutorials, and AI news",
    "url": "https://hoffdesk.com/blog",
    "author": "Matt Hoffman"
}

For base.html.j2:

page_title  # string — for <title> tag (falls back to site.title)
page_description  # string — for meta description
canonical_url  # string — full URL for this page
og_image  # string — full URL to hero image (or default)

For single post object:

post = {
    "title": str,
    "slug": str,
    "content": str,  # full HTML from Markdown
    "excerpt": str,
    "category": str,
    "tags": [str],
    "published_at": datetime,
    "updated_at": datetime,  # optional
    "author": str,
    "read_time": int,  # minutes
    "hero_image": str or None,  # relative path like "images/hero.svg"
    "hero_alt": str or None
}

For posts list (index/category/tag pages):

posts = [post, post, ...]  # paginated? up to you

# Plus pagination if you want it:
page_current = int
page_total = int
has_next = bool
has_prev = bool

For category/tag templates:

category  # string — "homelab", "openclaw", "ai-news"
tag  # string — the tag being filtered
post_count  # int — how many posts match

For feed.xml.j2:

build_date  # datetime — now
posts  # all published posts, newest first, full content

Extra: Image Pipeline (New)

I created image-pipeline-spec.md — read it. Key points:

1. Hero images live in data/blog/posts/{slug}/images/
2. Copy them to dist/blog/{slug}/images/ during build
3. Rewrite paths in generated HTML: images/hero.svg → /blog/{slug}/images/hero.svg

Example:

<!-- In template: -->
{% if post.hero_image %}
  <figure class="hero-image">
    <img src="{{ post.hero_image }}" alt="{{ post.hero_alt }}"
         width="1200" height="630" loading="eager">
  </figure>
{% endif %}

<!-- Generated HTML: -->
<img src="/blog/the-night-i-broke-dns/images/hero.svg" ...>

Green Light

Start building builder.py. Templates are frozen, context spec is above, image pipeline is documented.

When you have generated HTML, ping me — I'll verify visually and catch any CSS tweaks needed.

🎨

This response is also in workspace-socrates/ if you prefer reading there.