DID reference

Every DID form, every encoding, every edge case — as a lookup table.

This is the dictionary. For the reasoning, see Identity concepts.

Grammar

did:svrp:[<tag>:]<hex>

Part	Values
`tag`	`n` node · `o` operator · `h` host. Absent on the full-hex form.
`hex`	8, 10 or 64 lowercase hex characters.

^did:svrp:(?:([a-z0-9]):)?([0-9a-f]+)$

The three forms

Form	Length	Derivation	Where it appears
Short	8 hex	`SHA-256(key)[0:4]`	Bindings, operator records, `/peers`
Collision fallback	10 hex	`SHA-256(key)[0:5]`	Reserved; issued only on an 8-hex collision
Full-hex	64 hex	the 32-byte key itself	`/nodes`, running daemons

警告

Handle all three. An implementation that hard-codes only the 4-byte formula will reject a legitimate identity the first time a 10-hex one is issued — and will do so with an error that looks like forgery.

A tagged 64-hex value (did:svrp:n:df9d…) is not a shape this protocol produces. Reject it rather than guessing which half is meaningful.

Public key encodings

Endpoint	Encoding	Example
`GET /nodes`	lowercase hex, 64 chars	`df9d4b9f…02d`
`GET /peers`	base58 (Bitcoin alphabet)	`7441yUYc1qmW…`
Binding payloads	base58	`7441yUYc1qmW…`

Both are the same 32 bytes. Decode before hashing:

typescript

const key = /^[0-9a-f]{64}$/.test(publicKey)
  ? Buffer.from(publicKey, "hex")
  : base58.decode(publicKey);
if (key.length !== 32) throw new Error("not an Ed25519 public key");

注意

Base58 drops leading zero bytes. A correct decoder restores one zero byte per leading 1 in the string — otherwise a key beginning with 0x00 decodes to 31 bytes and fails.

Verification

form == full-hex   →  body == hex(key)
form == short      →  body == hex(sha256(key))[0:8]
form == fallback   →  body == hex(sha256(key))[0:10]

typescript

import { verifyIdBinding, parseDid, sameIdentity, fingerprint } from "libertynet-sdk";

verifyIdBinding(did, publicKey);      // boolean
parseDid(did);                        // { did, form, tag, body } | null
sameIdentity(didA, didB, publicKey);  // same node under two spellings?
fingerprint(publicKey);               // "268d:4fe0:b6ef:c390"

python

from libertynet import verify_id_binding, parse_did, same_identity, fingerprint

verify_id_binding(did, public_key)      # bool
parse_did(did)                          # ParsedDid | None
same_identity(did_a, did_b, public_key) # same node under two spellings?
fingerprint(public_key)                 # "268d:4fe0:b6ef:c390"

Comparison rules

Question	Correct test
Same node?	`sameIdentity(a, b, key)` — never `a === b`
Is this DID legitimate for this key?	`verifyIdBinding(did, key)`
Canonical short form for a key	`didFromPublicKey(key, "n")`
Show a human, for comparison	`fingerprint(key)`

警告

did:svrp:n:8545027b and did:svrp:df9d4b9f…02d can be the same node. String comparison will split one node into two across your database, dashboard and metrics — and the bug is invisible until someone notices a duplicate.

Fingerprints

fingerprint(key) = hex(sha256(key))[0:16] grouped in 4s
                 → "268d:4fe0:b6ef:c390"

For humans to compare out-of-band — on two screens, or read aloud. Nobody can diff 64 hex characters reliably; four groups of four is a task a person can actually do.

Worked examples

Real records from the live registry, 2026-07-31.

DID         did:svrp:df9d4b9f390bc49b2210eca81ca3c54c4d63b37b739442a21e28f2ed5b8aa02d
key (hex)   df9d4b9f390bc49b2210eca81ca3c54c4d63b37b739442a21e28f2ed5b8aa02d
form        full-hex        body == hex(key)                  ✓
short form  did:svrp:n:8545027b   (= sha256(key)[0:4])
fingerprint 8545:027b:6100:1591

DID         did:svrp:n:268d4fe0
key (b58)   7441yUYc1qmWVkAAUrPapKC13MXusEqDvvQJ1Pw5NNfg
form        short           body == sha256(key)[0:8]          ✓
fingerprint 268d:4fe0:b6ef:c390

Verify the whole network yourself

bash

node -e '
const { createHash } = require("node:crypto");
fetch("https://registry.libertynet.ai/nodes").then(r => r.json()).then(({nodes}) => {
  const B58="123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz";
  const b58=s=>{let n=0n;for(const c of s){const i=B58.indexOf(c);if(i<0)return null;n=n*58n+BigInt(i)}
    let h=n.toString(16);if(h.length%2)h="0"+h;let l=0;for(const c of s){if(c==="1")l++;else break}
    return Buffer.concat([Buffer.alloc(l),Buffer.from(h,"hex")])};
  const ok = nodes.filter(n => {
    const k = /^[0-9a-f]{64}$/.test(n.public_key) ? Buffer.from(n.public_key,"hex") : b58(n.public_key);
    const b = n.did.replace(/^did:svrp:(?:[a-z0-9]:)?/, "");
    return b.length===64 ? b===k.toString("hex")
         : b===createHash("sha256").update(k).digest("hex").slice(0,b.length);
  });
  console.log(`${ok.length}/${nodes.length} verified`);
});'
# → 27/27 verified

Why any of this

Identity concepts — what self-certifying buys you, and the verification order.

DID reference

#Grammar

#The three forms

#Public key encodings

#Verification

#Comparison rules

#Fingerprints

#Worked examples

#Verify the whole network yourself