Enhancement from the embedded-key security discussion (2026-07-08):
Proof's snippet key is already stronger than the Plausible/Umami model (their ingest endpoints are unauthenticated — public site identifiers only; Umami's data-domains guard is client-side in the tracker and trivially bypassed). The next step that would make Proof's anti-junk story the strongest in the category while keeping one-tag embeds:
🤖 Generated with Claude Code
Enhancement from the embedded-key security discussion (2026-07-08):
Proof's snippet key is already stronger than the Plausible/Umami model (their ingest endpoints are unauthenticated — public site identifiers only; Umami's
data-domainsguard is client-side in the tracker and trivially bypassed). The next step that would make Proof's anti-junk story the strongest in the category while keeping one-tag embeds:keysgains an optionalallowed_originscolumn (comma-separated origins, set at mint time via an optional field on both mint paths)keyAuth(or a follow-on middleware) rejects statements when the key has an allowlist and the request'sOriginheader is present-and-not-listed (absent Origin — curl/server-side senders — policy decision: reject when allowlist set, since the allowlist signals browser-only intent)🤖 Generated with Claude Code