Module encoding 

Human-friendly “prefixed” encoding for the public data types.

This module sits on top of the canonical byte / hex encoding in crate::types. It does not change a single bit of what is hashed, signed or verified — it is a pure presentation layer whose only job is to make a copy-pasted value:

• self-describing — a short tag (pk, sk, ki, blsag) up front says what kind of value it is, so a public key pasted where a key image was expected is caught immediately;
• typo-resistant — a trailing checksum detects the overwhelming majority of single-character mistakes, transpositions and truncated pastes before the bytes ever reach the cryptographic core.

Wire shape

  pk_3f8a…e1c0_d4e9a1b7
  │  │         │
  │  │         └ checksum: 4 bytes, hex (8 chars)
  │  └ body: the canonical hex encoding, exactly as `to_hex()` emits it
  └ tag: pk | sk | ki | blsag

Three _-separated parts. None of the parts can itself contain a _ (the tags are fixed, the other two are hexadecimal), so splitting on _ is unambiguous.

Checksum

checksum = BLAKE3(DOMAIN || tag || 0x00 || payload)[..4].

The tag is folded into the checksum pre-image (with a 0x00 separator so no tag can be confused with a prefix of another), which is what makes relabelling detectable: take a valid pk_… string, rewrite the tag to ki_, and the checksum no longer matches. So one check covers both “is the prefix coherent?” and “was the value mistyped?”.

This checksum is not a security primitive. An attacker can trivially compute a valid checksum for any bytes they like; its only purpose is to catch honest mistakes. Authenticity comes from the BLSAG proof, never from this.

Enums

Tag

Which kind of value a prefixed string carries. The string form is the human-readable prefix; it is also folded into the checksum.

Functions

decode_prefixed

Decode a tag_<hexbody>_<hexchecksum> string, verifying both the tag and the checksum, and return the raw payload bytes.

encode_prefixed

Encode payload as tag_<hexbody>_<hexchecksum>.