Documentation

1. Add Siere to your Vercel project

Find Siere on the Vercel Marketplace and click Add Integration. Select the project you want to connect.

This automatically sets SIERE_API_KEY and SIERE_SITE_ID on your Vercel project — no manual configuration needed.

2. Install the edge package

3. Add the middleware

Create or update your middleware file. AI agent requests get optimized content; human traffic passes through unchanged.

// middleware.ts
import { createSiereMiddleware } from '@siere.ai/edge-vercel';

const siere = createSiereMiddleware();

export default async function middleware(request: Request) {
  const response = await siere(request);
  if (response) return response;
  // ... your existing middleware logic
}

No config needed — the middleware reads SIERE_API_KEY from your environment automatically.

4. Deploy and verify

Push your changes and check the health endpoint:

curl https://your-site.vercel.app/__aeo/health

Then test with an AI agent user-agent:

curl -H "User-Agent: GPTBot/1.0" https://your-site.vercel.app/
# Should return structured markdown instead of HTML

1. Get your API key

Sign up at siere.ai and create a site in your dashboard. Copy your API key from the setup page.

2. Install the package

3. Set environment variables

# .env.local
SIERE_API_KEY=your_api_key_here

4. Add the middleware

// middleware.ts
import { createSiereMiddleware } from '@siere.ai/edge-vercel';

const siere = createSiereMiddleware();

export default async function middleware(request: Request) {
  const response = await siere(request);
  if (response) return response;
  // ... your existing middleware logic
}

Alternative: Proxy Mode

If you want Siere to run as a standalone edge function in front of your site (instead of as middleware), use proxy mode. Set SIERE_ORIGIN_URL to your site's URL.

// .env.local
SIERE_API_KEY=your_api_key_here
SIERE_ORIGIN_URL=https://your-site.com

// api/siere/route.ts
import { createSiereEdge } from '@siere.ai/edge-vercel';

export const GET = createSiereEdge();
export const POST = createSiereEdge();

5. Deploy and verify

curl https://your-site.vercel.app/__aeo/health

Test with an AI agent:

curl -H "User-Agent: GPTBot/1.0" https://your-site.vercel.app/
# Should return structured markdown instead of HTML

1. Pass config to the middleware

Replace the bare createSiereMiddleware() call with a config object. The middleware reads your sitemap.xml once per cache window and produces a markdown index agents can consume.

// middleware.ts
import { createSiereMiddleware } from '@siere.ai/edge-vercel';

const siere = createSiereMiddleware({
  canonicalOrigin: 'https://your-site.com',
  llmsTxt: {
    fallback: true,
    // Optional: hand-curated entries take priority over sitemap discovery
    entries: [
      { category: 'Pages', title: 'Home', url: '/', description: 'Site overview' },
    ],
    // Optional: enrich sitemap entries with each page's <title>+<meta description>
    // enrichFromHtml: true,
  },
  acceptHeader: { enabled: true },
});

canonicalOrigin is required for the fallback to locate your sitemap. The default sitemap path is <canonicalOrigin>/sitemap.xml — override with llmsTxt.sitemapUrl if yours lives elsewhere.

2. Verify the agent surfaces

# A markdown index built from your sitemap
curl https://your-site.com/llms.txt

# Same index plus extracted body text per page (token-capped)
curl https://your-site.com/llms-full.txt

# Health endpoint — confirm llmsTxtMode: "fallback"
curl https://your-site.com/__aeo/health

3. Optional: JSON-LD and well-known files

Siere also ships pure helpers for JSON-LD structured data (Organization, WebSite, SoftwareApplication, FAQPage, BreadcrumbList) and the well-known files /.well-known/security.txt, /ai.txt, and /humans.txt. These are framework-specific (you decide where to render the scripts and which routes to expose). Talk to us if you want a worked example for your stack.