Every item on the site is the unedited folder one recipe run published: a story, a playable game, a
comic, a song, a narrated short, a podcast episode. Each one links back to the recipe that made it
and to the Cronsole template that schedules it.
Nothing here is hand-written. What you see is what the runner wrote on its first live run, so the
rough edges are left in on purpose.
How it's built
site/ what GitHub Pages serves
├── index.html the viewer, copied from ai-automation-recipes/viewer/
├── catalog.js generated: every artifact, its media and its rendered markdown
└── <collection>/<YYYY-MM>/<date>_<slug>/ one folder per run, with its artifact.json
scripts/check_public.py the recipes repo's leak scan, run over this repo's history on every push
To add a run: copy its folder into site/ under the same layout, then rebuild the catalog from a
recipes checkout:
python scripts/build_viewer.py --output-dir ../ai-automation-recipes-demo/site \
--title "AI Recipes: Live Samples" \
--repo-url https://github.com/ai-automation-tools/ai-automation-recipes
Run python scripts/check_public.py here before you commit, and keep each file under GitHub's
100 MB limit. Pushing to main deploys.
License
MIT.
