Liquid Network

Liquid is a federated sidechain pegged to Bitcoin, developed by Blockstream. Funds are locked on Bitcoin via federation multisig and unlocked on Liquid as L-BTC.

Codebase: Elements (github.com/ElementsProject/elements), a Bitcoin Core fork with extensions.

Architecture

• Federation: ~15 functionaries (Blockstream + partners).
• Block production: round-robin among functionaries.
• 2-minute target block time.
• Peg-in: send BTC to federation multisig → equivalent L-BTC appears on Liquid after 102 confirmations.
• Peg-out: send L-BTC to federation → BTC released on Bitcoin after withdrawal queue.

Confidential Transactions

Liquid implements Confidential Transactions (CT):

• Output amounts are hidden via Pedersen commitments.
• Range proofs prove amount ≥ 0 and within valid range.
• Asset IDs also blinded (via blinded amounts + blinded asset commitments).
• Public key recipients can compute shared blinding factor and see the amount.

Issued Assets (IA)

Beyond L-BTC, Liquid supports:

• L-USDt (Tether on Liquid).
• L-CAD, L-EUR, L-CNY (some jurisdiction-specific).
• Custom asset issuance: anyone with funds can issue an asset.

Issuance specifies:

• Total supply (or issuance ongoing via reissuance tokens).
• Asset metadata (description, contract hash).
• Asset ID = hash of issuance tx + entropy.

Peg-in / Peg-out

Peg-in (BTC → L-BTC)

1. User sends BTC to federation multisig with claim script.
2. Wait 102 confirmations on Bitcoin.
3. User submits claim on Liquid; L-BTC issued.

Peg-out (L-BTC → BTC)

1. User submits L-BTC to federation address on Liquid.
2. Federation processes withdrawal queue.
3. BTC released to user's specified address on Bitcoin.

Peg-out is whitelist-based (regulated; Liquid-Pro members can peg out; retail users typically use exchanges for peg-out).

Implementation

• elementsd — full node (similar to bitcoind).
• liquid-network/elements — codebase, builds on Bitcoin Core.
• Wallets:
  • Blockstream Green — mobile wallet with Liquid support.
  • Sparrow — desktop, Liquid support.
  • Aqua — multi-asset Liquid wallet.

Use cases

• Stablecoin payments with privacy (L-USDt).
• Faster settlement between Liquid-aware exchanges (2 min vs 60 min).
• Asset issuance: STOs, regulated tokens, gift cards.
• Confidential commercial flows: amounts hidden from competitors.

Federation governance

• Functionaries hold multisig keys.
• Block-signing rotation among them.
• Transparent membership; replacements are governance events.

Lightning on Liquid

Liquid has Lightning compatibility:

• Lightning on Liquid (LBTC channels).
• Submarine swaps BTC ↔ L-BTC. Boltz, long the main public provider, is offline since 2026-08-03: all mainchain, Lightning and Liquid swaps disabled "until further notice", no relaunch date announced as of 15 September 2026.
• Blockstream Swaps — in-house replacement covering the same BTC ↔ L-BTC ↔ Lightning routes (submarine, reverse submarine and chain swaps). Limited beta since 10 Aug 2026, "in beta testing with select participants"; access by request.

Compared to other L2s

Security: the September 2026 cache-collision exploit

On 6 September 2026 (Liquid block 4,050,336) an attacker exploited a cache-key collision in Elements' range-proof verification cache. Elements memoises positive range-proof verifications; a cache hit is a positive result, so a key collision bypasses verification entirely.

Two defects are visible in the Elements history:

• Releases up to and including elements-23.3.3 (13 Apr 2026) keyed on salted_hash(proof || commitment) only, omitting the asset generator and the scriptPubKey.
• The commit binding those in (212c43f475, on the elements-23.3.x branch 3 Sept 2026) concatenated the fields without length prefixes, so bytes shift across the boundary between the two variable-length fields and distinct tuples collide.

Outcome, as of 15 September 2026:

• ~3,998.5 unbacked L-BTC minted and pegged out for 4,000 BTC ($320M); the federation reserve fell from ~4,200 BTC to ~197 BTC (~95% drained).
• No key compromise. Per Blockstream's status page the funds moved via the SideSwap peg-out authorisation key, "but that key was not compromised, nor were any others". The HSMs signed because the peg-out was valid under the consensus rules they ran — an HSM policy is only as sound as the consensus code behind it.
• Bridge nodes disabled 7 Sept 2026; block production resumed transaction-free 10 Sept 2026; peg operations (including PAK-authorised peg-outs) stayed suspended past that.
• Attackers returned 3,400 BTC, kept 598.5 BTC ($47M) as a self-declared bounty; Blockstream refused to pay it.

Minimum safe version: elements-23.3.4 (9 Sept 2026). PR #1600 "sigcache: harden range proof cache keys and add -norangeproofcache option" moves both proof caches from raw CSHA256 concatenation to CHashWriter (length-prefixed serialisation), adds the missing vTags to the surjection-proof key, and adds -norangeproofcache to disable the range-proof cache without recompiling.

Limitations

• Federation trust: 11-of-15 multisig; large compromise = funds loss. Consensus-code bugs are the other systemic risk, and the one that actually fired in September 2026 (above).
• Peg-out gated for retail users (typically via exchanges).
• Smaller ecosystem than EVM L2s.

See also

• rgb/SKILL.md
• taproot-assets/SKILL.md
• statechains/SKILL.md
• ../../infrastructure/btcpay/SKILL.md