This is a static site with no package or build dependency.

From the repository root:

python3 -m http.server 4173

Then open http://localhost:4173/website/.

Run the dependency-free smoke check after changing links, assets, or sections:

python3 scripts/check-website.py

After changed pages are live, preview and send their canonical URLs to IndexNow:

python3 scripts/submit-indexnow.py --dry-run \
  https://www.semper.systems/about.html
python3 scripts/submit-indexnow.py \
  https://www.semper.systems/about.html

Omit positional URLs to submit every URL in website/sitemap.xml.

An HTTP 200 or 202 response confirms receipt, not crawling, indexing, ranking, or recommendation. The sitemap remains submitted separately in Google Search Console.

Source, contribution, guide, and issue links point to https://github.com/niharnm/Semper.

website.hero-repository-cta-copy.v1 assigns visitors to one of two labels:

• control: View the source
• treatment: Open Semper on GitHub

The assignment is stored in localStorage, with sessionStorage as a fallback. Force a variant for local testing without changing the stored assignment:

http://localhost:4173/website/?semper_ab=website.hero-repository-cta-copy.v1:control
http://localhost:4173/website/?semper_ab=website.hero-repository-cta-copy.v1:treatment

The page dispatches semper:experiment events on window. Event details contain anonymous subject and session IDs, the experiment and variant, and either experiment_exposure or experiment_outcome. Events stay in the browser; the site does not send them over the network, so this repository does not yet aggregate results or select a winner.

Run the dependency-free tests from the repository root:

node --test website/ab-testing.test.js

The homepage distinguishes the published Sound release from development modules. When a packaged release changes, update the download context, SoftwareApplication.softwareVersion, FAQ text and schema, about.html, llms.txt, and the modified pages in sitemap.xml together. A merged feature PR alone is not evidence of availability in the download.

The comparison table links to primary feature documentation. Its dated feature descriptions are separate from the performance section. Do not use feature counts, missing measurements, or an observational sample as a ranking.

For benchmark publication, consume the reviewed evidence from scripts/benchmarks/evidence/ only after that work is integrated. Schema v1 uses status and comparison_eligible: unmeasured has no metrics; observational is not eligible for comparisons. Keep the current unmeasured presentation until controlled evidence is approved. Add immutable report and raw-data links, exact build and workload details, units, repetitions, and variation with any published result. Do not automatically render arbitrary JSON values as scores.

Verification:

python3 scripts/check-website.py
python3 scripts/test_website_tools.py
node --test website/ab-testing.test.js

Check the homepage at desktop and narrow mobile widths. Exercise module scope disclosures by keyboard, scroll the comparison table to its final column, try the Sound sliders and mute buttons, and inspect reduced motion.