REFINERY Website

The public site for The REFINERY, a nonprofit robotics makerspace affiliated with BioNanomics. Built with Astro and Tailwind CSS; content lives in Markdown/MDX.

See plan.md for the original project brief and CONTRIBUTING.md for how to add a news story, team, program, or event.

Development

npm install
npm run dev       # http://localhost:4321

Commands

Command	Action
npm install	Install dependencies
npm run dev	Start the local dev server
npm run build	Build the production site to ./dist/
npm run preview	Preview the production build locally
npm run astro check	Type-check the project

Updating the impact stats

The impact numbers shown on the homepage and in the About page's Impact section live in one place: src/data/impact.ts. Both pages render <ImpactStats />, which reads that file, so editing it updates both and they can't drift apart.

Edit the value of the clause you want to change and redeploy — no other code changes needed. The numbers are also repeated in prose in public/llms.txt, which is not generated from impact.ts, so update that too if a figure changes.

Deployment

Currently in review mode, not launched. The repo is public and GitHub Pages (Source: GitHub Actions) serves the site at https://bionanomics.github.io/REFINERY-site/ so it can be reviewed before the real launch. site in astro.config.mjs is temporarily https://bionanomics.github.io with base: '/REFINERY-site', public/robots.txt disallows all crawling, and SITE_INDEXABLE in src/layouts/BaseHead.astro is false (forces noindex on every page). public/CNAME is still deliberately absent, so nothing claims refineryrobotics.org — the repo's prior private visibility was the only thing protecting it from being publicly reachable, and that protection is now gone in exchange for a link anyone can view without a GitHub account.

Reverting to the pre-review state (e.g. if review needs to pause): re-run gh repo edit --visibility private, restore Allow: / + the Sitemap: line in public/robots.txt, and drop the base/site review-mode change in astro.config.mjs back to https://refineryrobotics.org with no base.

Launch order, when ready — this replaces review mode entirely: flip SITE_INDEXABLE to true, restore astro.config.mjs's site/base to the apex-domain values above, and restore public/robots.txt to Allow: / plus its Sitemap: line. Then work through the domain steps below. site in astro.config.mjs is https://refineryrobotics.org and DNS already points there via Cloudflare, but public/CNAME is deliberately absent, so nothing claims the domain until step 1.

Launch order, when ready. Step 1 is the arming step: GitHub Pages reads CNAME from the deployed artifact and claims the custom domain from it, so don't enable Pages before it.

Because step 1 arms launch, keep it out of any branch carrying unrelated work. Merging such a branch would go live early, before the redirect and headers below are in place. Commit CNAME on its own, when you are ready to work through steps 2–8 in one sitting.

Step 4 is the one step worth doing before step 1 rather than after. www and the apex both resolve today, so the moment the site serves content both hostnames serve identical pages with identical canonical tags, and canonicals are hints rather than directives. Redirecting first means Google never sees two live hostnames; redirecting after means a possible change-of-address and weeks of recovery. Nothing breaks if the redirect exists before there is a site to redirect to.

1. Create public/CNAME containing refineryrobotics.org, and merge to main.
2. Settings → Pages → Source: GitHub Actions. Not "Deploy from a branch" — that runs Jekyll against the repo root, which has no built HTML (dist/ is gitignored), and it doesn't satisfy actions/deploy-pages, so the workflow keeps failing. This is an easy trap to fall into; it has already caught us once.
3. Confirm the custom domain registered, then enable Enforce HTTPS.
4. In Cloudflare, handle both redirects. Safe to do before step 1; see the note above.
   • 301 www.refineryrobotics.org/* → https://refineryrobotics.org/$1, preserving path and query (both hostnames currently resolve, and neither redirects).
   • SSL/TLS → Edge Certificates → enable Always Use HTTPS. Plain http:// currently answers directly with no redirect. HSTS in step 5 only protects repeat visitors; this is what covers first contact.
5. In Cloudflare, add security response headers via Transform Rules → Modify Response Header. GitHub Pages can't set these, so Cloudflare is the only place they can come from. Do this after step 3 — HSTS before a working certificate locks visitors out:
   • Strict-Transport-Security: max-age=86400; includeSubDomains to begin with. Raise it toward 31536000 once you're confident HTTPS is stable: the value is cached by browsers, so a long max-age set too early is hard to walk back. Leave preload off entirely; getting off the preload list is much harder still.
   • X-Content-Type-Options: nosniff
   • Referrer-Policy: strict-origin-when-cross-origin
   • Permissions-Policy: geolocation=(), camera=(), microphone=()
6. Verify /robots.txt, /llms.txt, and /sitemap-index.xml serve 200 at the domain root, and that a few canonical URLs resolve 200. Cloudflare appends its own content-signals block to robots.txt, so confirm the Sitemap: line survives the merge.
7. Verify refineryrobotics.org as a domain property in Search Console, and submit the sitemap.
8. Run the Rich Results Test against every page that emits structured data. This needs a live URL, so it is the one check that can't be done before launch.
   • /: Organization, WebSite
   • /about/: Organization with the facility Place, plus one Person per bio
   • /about/teams/: BreadcrumbList
   • /news/re-blitz-summer-build-kickoff/: Article, BreadcrumbList
   • /programs-events/monster-match/: Event, BreadcrumbList

site in astro.config.mjs and public/CNAME must always agree.