architecture & protocol

Docs

A sovereign proof-of-work UTXO chain — Bitcoin's trust model, a light client that lets a phone verify it, and an honest account of where it stops.

consensus

Proof-of-work, most-work wins.

Every block carries a real 256-bit target; a block is valid only if doubleSHA256(header) ≤ target. Nodes follow the chain with the most accumulated work (summed 2²⁵⁶ / (target+1)) — never the longest by count, which timestamp games can cheat. Each node re-validates every block from the pinned genesis and trusts no one.

2 min
target block time
2016
retarget interval (blocks)
100
coinbase maturity
240-bit
powLimit (easiest target)
money

Fixed supply, honest issuance.

keys & transactions

secp256k1, domain-separated signing.

A transaction spends UTXOs; each input carries a witness [pubkey, signature]. Signatures are ECDSA over SHA-256("WAKECHAIN_TX_V1|" + genesis + "|" + tx) — the network domain binds every signature to this chain, so a transaction can't be replayed onto a fork. Addresses are SHA-256(pubkey)[:40]. Keys are generated and signed on the user's device; the node never sees a private key.

light client

Your phone verifies the chain — trusting no full node.

Each block commits a single 32-byte root over the whole UTXO set — a sparse-Merkle accumulator — inside its coinbase. A light client holding only PoW-verified headers can then check "does this coin exist?" from the root and an O(log n) proof alone. A lying node cannot forge a proof that reconciles to the committed root, and non-membership is provable too. This is what powers /verify and the wallet's proven balance.

Honest note: the committed root lives in the tip's coinbase and commits the state as of the previous block (H−1). So a just-received coin shows one block later. The wallet labels this plainly ("proven @ block H−1"). A post-state-root commitment is the planned protocol upgrade.

defense in depth

Guardians, fenced. Borrowed hashpower.

The guardian layer

An optional M-of-N recovery + anti-scam layer with post-quantum (hash-based) co-signatures. It is bounded by consensus so it can protect without becoming the third party it set out to remove — it can freeze or recover under rule, never silently seize.

Merged mining

An AuxPoW path lets WakeChain inherit a parent chain's hashpower — solving the cold-start miner problem while staying sovereign proof-of-work. Designed and tested; not yet active against a live parent.

private messaging

WakeChat — end-to-end encrypted, with forward secrecy.

Your chat identity is a secp256k1 key generated on your device — no phone, no email, no name; your address is the public key. Messages are sealed with a Double Ratchet (the Signal construction, rebuilt from scratch here): an ECDH handshake on secp256k1 seeds an HKDF-SHA256 root chain; each message then draws its own key from an HMAC-SHA256 symmetric chain and encrypts with AES-256-GCM, the message header bound in as associated data. A DH ratchet injects fresh entropy every time the direction of the conversation flips, and consumed keys are destroyed.

per-msg
key, thrown away after one use
FS
forward secrecy — a stolen key can't read the past
PCR
post-compromise recovery — one round-trip re-secures
0
plaintext the relay can see — ciphertext only

Transport is an off-chain, store-and-forward relay on the same nodes: it holds ciphertext only (7-day TTL), gossips between nodes, and verifies each sender signature before storing — so a forged or tampered envelope is rejected before it is ever trusted. Keys, history and contacts live encrypted on your device.

Honest note: the ratchet was built, found broken by two independent internal reviews, redesigned, and re-audited — it is self-audited twice, not yet third-party audited, so we do not call it "perfect." Forward secrecy protects what the network holds, not plaintext history saved on a device someone fully controls. The relay still sees metadata — which address talked to which, and when (not what) — and your IP, as with every app; use a VPN or Tor to hide it. A rare simultaneous-first-message race can drop an opening message (resend resolves it). The full story is in the whitepaper §8.

honest limits

What it is not, yet.

Two nodes and one miner, all founder-operated — so the "two independent nodes agree" check the wallet performs is only as strong as their independence, which today is limited. A single CPU miner means the chain is not yet economically secure against a 51% attack (merged mining is the intended answer). This is a young mainnet — hardened and honest, but NOT yet economically secure. It holds real value, which makes the limits above matter more, not less. Every page here says exactly what it proves and where it stops.