tesora-www

Marketing + content site for Tesora AI. Specialty P&C actuarial pricing.

Built for static generation (Astro 6) so search crawlers and LLM ingest agents see real HTML, not an empty <div id="root">. Voice flows from ~/tesora-harness/shared/prompts/voice.ts — the harness is the source of truth for tone.

Stack

• Astro 6 + TypeScript (strict) + Bun
• Tailwind v4 (@tailwindcss/vite)
• MDX for content collections
• @astrojs/sitemap for auto sitemap.xml

Commands

bun install
bun dev           # localhost:4321
bun build         # static output to ./dist/
bun preview       # serve ./dist/ locally
bun astro check   # typecheck

Layout

public/                 robots.txt, llms.txt, favicons, OG images
src/
  components/SEO.astro  per-page metadata: title, og:*, twitter:*, JSON-LD
  layouts/Layout.astro  base HTML shell
  styles/global.css     Tailwind import + design tokens
  pages/                static + dynamic routes
  content/              typed content collections (planned)
  data/                 manifests for programmatic page generation (planned)
  scripts/              one-shot generators (planned)

SEO + GEO posture

• Zero-JS-by-default static HTML. Crawlers and LLM agents see content directly.
• Every page has full metadata + JSON-LD (Organization + page-specific schema).
• /llms.txt for LLM discovery, /robots.txt allows GPTBot, ClaudeBot, PerplexityBot, etc. explicitly.
• Programmatic sub-page generation drives the long-tail. Pages target queries like:
  • "specialty P&C actuarial pricing for [line of business]"
  • "exposure base [type] [audience]"
  • "[insurance niche] pricing tools"
• Internal linking: every long-tail page links to 3-5 siblings and back to hub pages.

Status

Foundation only. DESIGN.md and the long-tail content manifest are next. See CLAUDE.md for detailed project context.