A standard.site validator. Paste a webpage URL, see its site.standard.* records and whether the chain from page -> PDS -> .well-known verifies. Built with Hono on Cloudflare Workers.

Checks #

Check	What it verifies
document link tag present	page declares `<link rel="site.standard.document">`; N/A on publication-root URLs
document AT-URI is well-formed	link href parses as `at://{did}/{collection}/{rkey}`
document AT-URI collection	collection segment is `site.standard.document`
document record key is valid TID	rkey is a real ATProto TID
resolve document PDS	document's DID resolves to a PDS via plc.directory or `did:web` `.well-known/did.json`
publication AT-URI is well-formed	same shape check; source can be `document.value.site`, a page `<link>` tag, or `.well-known` discovery
publication AT-URI collection	collection segment is `site.standard.publication`
publication record key is valid TID	rkey is a real ATProto TID
resolve publication PDS	publication's DID resolves to a PDS
document and publication share DID	both records live on the same identity
well-known round-trip	`/.well-known/site.standard.publication[/path]` body equals the resolved publication AT-URI (path appended for subpath publications)
canonical URL match	`publication.url + document.path` equals the inspected URL (trailing slashes normalized)
publication required fields	`url` and `name` are present per the lexicon
document required fields	`site`, `title`, `publishedAt` are present per the lexicon

TODO #

flag any unknown/wrong standard.site lexicons.
Caching. Can use KV with expiration perhaps, and specific cache skipping.
Cloudflare turnstile?

Develop #

bun install
bun run dev      # vite + cloudflare plugin
bun run check    # biome + tsc
bun run fix      # biome auto-fixes
bun run deploy   # vite build + wrangler deploy

AI usage disclaimer #

This project was made using Claude Code (plus some elbow grease, but it's surprising how little).

Clone this repository

is it standard.site? #

Checks #

TODO #

Develop #

AI usage disclaimer #