Is this a receipt at all?
The text must parse as JSON and contain a body and a signature block. Anything else stops here, before any cryptography runs.
Exactly which bytes were signed?
The body is rewritten in its RFC 8785 canonical form: sorted keys, no whitespace, one number format. Two people with the same receipt always get the same bytes.
Has anything changed?
SHA-256 of the canonical bytes must equal the digest in the receipt. Change one character of one field and this check fails. That is what "flip one character" shows.
Did the issuer sign it, classically?
The Ed25519 signature is checked against the issuer public key carried in the receipt. This is the leg every existing tool can already verify.
Did the issuer sign it, post-quantum?
The FIPS 204 signature is checked against the same issuer. Removing it is a downgrade attempt: "strip the PQ signature" shows the verifier refusing it.
Both legs valid, and a known issuer?
The hybrid suite requires both signatures, and the issuer keys in the receipt must be in the registry the verifier trusts: here, the keys issued in this tab; in production, the published key registry. Only then is the verdict VALID. The production profile also checks the chain and inclusion.