flnzba.github.io

Florian Zeba's personal site — built with Eleventy and styled with Tailwind CSS. Deployed to GitHub Pages at www.fzeba.com.

Stack

• Eleventy 3 static site generator
• Tailwind CSS 4 generated with the official CLI
• Nunjucks templates
• Pagefind for full-text search
• Crier (Python) for cross-posting to DEV.to and Hashnode via GitHub Actions

Local development

npm install
npm run serve              # http://localhost:8080
npm run build              # outputs _site/
npm run search             # build pagefind index after build

Authoring a new post

1. Create src/posts/<slug>/index.md. Use kebab-case for the slug — it becomes the URL: /posts/<slug>/.
2. Drop any cover/inline images alongside index.md (they're passthrough-copied).
3. Use this frontmatter template:

---
title: "Post title"
description: "1–2 sentence description (used as RSS summary and OG description)."
date: 2026-05-01
updated: 2026-05-01            # optional
tags: [data-engineering, python]
cover:
  src: ./cover.webp
  alt: "Cover image alt text"
draft: false
canonical_url: https://www.fzeba.com/posts/post-slug/
published: true                # set false to skip cross-posting
---

4. Commit and push to main — the deploy workflow builds + publishes the site, and the cross-post workflow posts to DEV.to and Hashnode (see below).

Authoring a project

Same shape, but in src/projects/<slug>/index.md. Projects are not cross-posted.

Image store

Paste standalone images into src/image-store/. On build, supported image files are copied to /image-store/<filename>, preserving subfolders. For example, src/image-store/report/chart.webp is available at https://www.fzeba.com/image-store/report/chart.webp.

Cross-posting (Crier)

Cross-posting is automated via .github/workflows/crosspost.yml. After a push to main that touches src/posts/**/*.md:

1. The workflow installs the pinned Crier version from requirements.txt.
2. scripts/prepare_crosspost.py finds newly added src/posts/*/index.md files and writes sanitized staging files under /tmp/flnzba-crosspost.
3. The staged files keep only Crier metadata in front matter; the API body is frontmatter-free Markdown with relative links resolved and indented code blocks fenced for DEV.to/Hashnode rendering.
4. Runs crier audit <staged-dir> --publish --batch --long-form against DEV.to and Hashnode.
5. Commits the updated SQLite registry .crier/crier.db back with [skip ci].

Required repo secrets

Secret	Format
DEVTO_API_KEY	DEV.to API key
HASHNODE_API_KEY	<token>:<publication_id>

Setting them

Either one secret at a time:

gh secret set DEVTO_API_KEY            --repo flnzba/flnzba.github.io
gh secret set HASHNODE_API_KEY         --repo flnzba/flnzba.github.io

…or in bulk from a local .env file:

cp .env.example .env       # fill in real values
npm run secrets:push       # uploads each KEY=VALUE as a repo Actions secret
gh secret list --repo flnzba/flnzba.github.io   # verify

.env is gitignored — it never reaches the repo. The helper script (scripts/sync-secrets.sh) only pushes the values into GitHub's encrypted secret store, which the workflow reads at run time via ${{ secrets.* }}.

Caveats

• canonical_url is required for SEO — it tells DEV.to and Hashnode that www.fzeba.com is the original source.
• The workflow only publishes newly added post files from the triggering commit or the optional manual post input, so enabling the registry does not publish the entire archive at once.
• The [skip ci] marker in the registry-update commit prevents an infinite loop with the deploy workflow.

License

Content © Florian Zeba. Code MIT.