# superschema.ai for AI agents

SuperSchema is an AI visibility platform: it generates, deploys, and monitors
Schema.org structured data so businesses get found and cited by AI systems.
This page is the machine quickstart. Facts below were verified 2026-08-19.

## The one thing agents get wrong here

There are FOUR sitemaps and nothing marks any of them canonical. They are
near-disjoint. Fetching only /sitemap.xml shows you 22% of the site and none
of the Discovery Network registry.

| sitemap | urls | contents |
|---|---|---|
| /sitemap.xml | 53 | marketing pages and tools |
| /ai-sitemap.xml | 19 | strict subset of /sitemap.xml, safe to skip |
| /docs/sitemap.xml | 72 | documentation knowledge base |
| /api/discovery/sitemap.xml | 116 | all /registry/* pages, zero overlap |

240 distinct URLs total across all four.

## Intent to surface map

| you want | fetch |
|---|---|
| what SuperSchema is, as markdown | /content/home.md |
| documentation index | /docs/llms.txt |
| any docs article as markdown | /docs/{slug}.md |
| marketing pages as markdown | /content/{slug}.md (home, pricing, aeo, geo, and more) |
| the public API, described | /openapi.json |
| API discovery catalog | /.well-known/api-catalog |
| a business in the registry, as JSON | /api/discovery/registry/{slug} |
| every registry slug | /api/discovery/slugs |
| agent skills for this site | /.well-known/agent-skills/index.json |
| everything in one file | /llms-full.txt |
| query this site over MCP | POST /mcp (streamable HTTP; card at /.well-known/mcp/server-card.json) |
| agent auth and rate limits | /auth.md |

## Structural gotchas

- Registry slugs encode domains with dots replaced by hyphens. Domain to
  slug is reliable; slug to domain is not recoverable. The JSON mirror
  returns the real domain.
- A missing /registry/{slug} page returns HTTP 200 with an 8,352 byte app
  shell. Test existence via /api/discovery/registry/{slug}, which 404s
  honestly.
- Unknown extensionless paths generally return 200 with the app shell.
  Do not treat a 200 alone as proof a page exists.
- Public API endpoints are limited to 100 requests per 15 minutes per IP.

## Content usage

Content signals are declared in /robots.txt: search=yes, ai-input=yes,
ai-train=yes. Attribution appreciated: link to the page you used.
