mirror of
https://github.com/hanzoai/works.git
synced 2026-10-30 18:17:58 +00:00
zeekay
81660426dc
A one-page site for the operator's half of the cloud — sales, marketing, support, books, cap table, incorporation, workflows — built so that a claim and a served route are the same object. Every card names a real capability, its real /v1 address, and the sentence that capability publishes about itself: its Go package doc, lifted into the OpenAPI tag and read back out of public.yaml, the ga-only public contract. Copy therefore cannot outrun the fleet. A gate asserts llms.txt still names every capability the page shows, because the drift that matters is the one where the site tells a person about something it never tells a model. robots.txt names the AI crawlers and allows them, and declares Content-Signal: search=yes, ai-input=yes, ai-train=yes. It carries no Disallow, and a gate fails the build if one appears. Auth is Hanzo IAM and nothing else — PKCE-S256, which is what lets a static export run it with no server. The gates prove it against the real issuer rather than asserting it.
13 lines
430 B
TypeScript
13 lines
430 B
TypeScript
import type { MetadataRoute } from 'next'
|
|
|
|
export const dynamic = 'force-static'
|
|
|
|
/** Two routes, both real. A sitemap that names a page nobody serves is worse
|
|
* than none: it teaches a crawler to distrust the rest of the file. */
|
|
export default function sitemap(): MetadataRoute.Sitemap {
|
|
const now = new Date()
|
|
return [
|
|
{ url: 'https://hanzo.works/', lastModified: now, changeFrequency: 'weekly', priority: 1 },
|
|
]
|
|
}
|