Wraith Protocol — Landing Page

The landing page for usewraith.xyz. A minimal, dark, developer-focused single-page site that explains what Wraith Protocol is and links to docs, demo, and console.

Stack

• Vite + React + TypeScript
• Tailwind CSS v4
• Deployed to Vercel

Development

pnpm install
pnpm dev

Build

pnpm build

The build pipeline performs the following steps:

1. Compiles TypeScript and builds the client production assets via Vite.
2. Runs the Open Graph image generator (scripts/og.ts) to pre-render cards for all routes and patches the output HTML files with page-specific titles, descriptions, and JSON-LD breadcrumb schemas.
3. Runs the sitemap generator (scripts/sitemap.ts) to generate a fresh sitemap.xml covering all static and dynamic routes.
4. Runs the RSS feed generator (scripts/gen-rss.mjs) to generate feed.xml and per-tag feeds under public/feed/tag/.

Sitemap and RSS generation

Both generators are deterministic: running them twice on the same source produces byte-identical output.

Sitemap — scripts/sitemap.ts is the single entrypoint for both the build pipeline and manual generation:

pnpm generate:sitemap   # regenerate public/sitemap.xml
pnpm check:sitemap      # fail if public/sitemap.xml is stale (used in CI)

lastmod dates are derived from content metadata (blog date frontmatter, case-study integrationDate) or the git commit that last touched each route's source files. Routes with no trustworthy date (tag archives, author pages) omit <lastmod> entirely rather than stamping today's date.

scripts/gen-sitemap.mjs has been retired. All callers should use scripts/sitemap.ts via tsx.

RSS — scripts/gen-rss.mjs generates public/feed.xml and per-tag feeds. The <lastBuildDate> is set to the publication date of the most-recent post, not the wall-clock time, so repeated builds on unchanged content produce the same file.

Serverless routes

vite preview only serves the static build, so the API routes (/api/subscribe in api/subscribe.ts and /api/og in src/api/og.tsx) are unavailable locally and on CI runners. pnpm preview:functions serves the same dist/ output with those routes mounted on port 4174:

pnpm build
pnpm preview:functions

Smoke tests

pnpm test:smoke

Smoke tests run against scripts/preview-server.ts over HTTP: the subscribe proxy (including validation, rate limiting, duplicate submissions, provider outages and the unconfigured case) and the OG image endpoint (rendering, query escaping, cache headers and the failure response). Buttondown is stubbed, so no mail is sent and no test touches the network. The suite runs on every pull request in CI.

The subscribe endpoint accepts JSON bodies up to 2,048 bytes and the documented newsletter (default) and careers tags. It allows five submissions per Vercel client IP per 15-minute window; the in-memory, HMAC-keyed limiter avoids retaining raw IP addresses and is scoped to each warm serverless instance.

Link checks

pnpm build && pnpm check:links                 # the build, served by scripts/preview-server.ts
pnpm check:links --base https://usewraith.xyz  # production
pnpm test:link-check                           # unit tests for the checker

scripts/check-links.ts starts from the sitemap and renders every public route in Chromium. It fails on routes that answer with an error or the 404 page, internal links to missing routes, files or #fragments, canonical and hreflang URLs that are missing, duplicated, off-site or not the page itself, feeds that are not served as XML, and external links. --report <file> writes the results as JSON and --no-external skips external links.

External links must point at a host listed in scripts/link-check/external-links.json. Add a host there when the site starts linking to it, and give a reason whenever an entry needs acceptStatus or skipRequest. Approved links are requested through scripts/link-check/safe-request.ts: only public addresses are allowed (loopback, private, link-local, IPv4-mapped forms of those, and so on are refused), each hop connects to the address its DNS lookup validated, and redirects are re-validated. The browser that renders pages can only load the site under test.

CI runs the checker on every pull request against the build, and every day against production.

SEO & Metadata

• robots.txt: Located in public/robots.txt and copied to the build root. It allows search engine crawling while excluding staging, preview, admin, and 404 routes, and points crawlers to the sitemap location.
• sitemap.xml: Dynamically compiled during build time into dist/sitemap.xml and public/sitemap.xml.
• JSON-LD: Embedded in the site's <head>:
  • Global Organization schema representing Wraith Protocol.
  • Global SoftwareApplication schema representing the Wraith SDK.
  • Page-specific BreadcrumbList schema dynamically injected for each static route slug (e.g. /stellar, /faq, etc.) at build time.

Format

pnpm format        # write
pnpm format:check  # check only