Cypher Box

📢 ⚠️ Attention Lightning Network node runners and operators: If onchain fees spike, Flood-and-loot becomes a systemic risk: Flood-and-loot is a theft attack on Lightning routing nodes. The attacker does not break any cryptography. They arrange things so that a routing node cannot defend its own money in time, and then they take it. How it works: Every Lightning payment that is mid-flight comes with a safety net. If something goes wrong, the node can fall back to the Bitcoin blockchain and reclaim its funds. But that safety net has a deadline. You have to get your transaction confirmed on-hain before a timer runs out. The attack turns that deadline into a trap. How it can be executed: - The attacker opens channels with many routing nodes and becomes their payment counterparty. - They push a lot of payments (HTLCs) through those nodes and leave them hanging, unsettled. - They force all of those channels to close onchain at the same moment. Now every victim needs a defensive transaction confirmed at once, all competing for the same limited block space. - Most of them miss the deadline. The attacker keeps whatever the victims could not defend. The unfair part is that the attacker does not have to win the bidding fee auction for block space. They only need their victims to lose it. When the blockchain is busy, that happens on its own. Why it is cheaper than it sounds: The attacker does not even have to create mempool congestion. They can wait for a moment when Bitcoin is already busy, like an Ordinals inscription wave or a bull-market hype, and strike then. The congestion is free cover. There is also a sharper version called replacement cycling. In that one, the attacker quietly keeps the victim's defensive transaction from ever reaching miners, so it simply never confirms. Why it will not go away: Most lightning network attacks are just vandalism. They cost the attacker money and return nothing, so people get bored and stop. But Flood-and-Loot is different, because it actually pays. The reward per channel is small, but across dozens of victims it adds up quickly. What it looks like from a wallet: If you use a custodial lightning wallet, where a company runs the node for you, you will mostly just see payments fail or get delayed. The company should absorb any real loss. But that company, if it runs its own in-house node, is itself an operator in exactly this sense. It holds liquidity and forwards HTLCs, so it is a target too, and the operator defenses below apply to it. If you run your own self-custodial / routing node, you carry the risk directly. A channel might close on its own. Your funds can sit pending on-chain for hours, and the closing fee can come out high. The only amount truly in danger is whatever was mid-payment at the time, not your whole balance. A simple way to read it: a lightning payment that failed is fine, just try again later. A channel force-closing while fees are high is the moment to pay attention. What to do: Operators, routing nodes (you are the target): - Use anchor channels, and keep spare on-chain coins set aside so you can always pay to close a channel in a hurry. - Limit the size of payments you forward, and how much you allow in-flight per channel. The loot is capped by what is in-flight, so smaller limits mean a smaller worst-case loss. - Give yourself more runway by widening your timelock settings. - Be cautious when a flood of channel requests arrives from strangers. - When the mempool is busy, treat it as a danger period and pull back. Users: Do not keep large savings on Lightning. It is a spending wallet. Let a well-run, updated app carry the small amounts, and keep the serious money in cold storage. LSP wallets: These wallets lean on a Lightning Service Provider. The LSP runs the big node and does the forwarding, so you sit as a leaf on the edge of the network. You do not forward strangers' payments, which means you are not who flood-and-loot is hunting. Your only real exposure is the single channel between you and the LSP, and modern LSP wallets handle a congested close well. Keep the app updated, keep real savings elsewhere, and know that you are trusting the LSP to stay online and healthy. The LSP itself is the real target. It holds liquidity and processes payments for thousands of users at once, so if it gets looted or forced into a mass close, everyone it serves feels it. The app in your pocket is a leaf, but the LSP behind it is a hub that has to defend itself like any routing operator. An LSP has to be careful about who it opens channels with. More precisely, it has to control how much money it ties up with peers it has not vetted: - Rate-limit channel opens, and charge for them. - Use just-in-time channels, tied to a real incoming payment, instead of handing out liquidity on request. - Cap how much can sit in-flight with any single peer. - Watch for a burst of channel requests from brand-new nodes. That is what a setup phase looks like. Ark wallets: Ark works differently. On the provider side: the bridge to Lightning runs on VHTLCs (virtual HTLCs that live inside the Ark construct), not funded channels, so there are no per-user channels to mass-force-close, and a user can always exit on their own even if the provider is offline. That is why Ark users are not flood-and-loot victims. An LSP is exposed to flood-and-loot by its own users, because those are funded channels. Ark's VHTLC model removes that vector entirely: your own users can't attack you. The Ark provider only has to worry about its own Lightning peers. So exposure sits one step back, at the real Lightning routing node that every Ark-to-Lightning bridge ends at. The catch: None of this fully fixes it. Flood-and-loot works because Lightning depends on winning a race against a blockchain that charges by the block, and that race is decided one layer below Lightning itself. Harden your nodes! If Bitcoin network fees spike, that's the Flood-and-loot starting gun, and by then it might be too late to prepare.

Here’s a moonshot idea to accelerate bitcoin payments adoption:

Sorry EU & UK plebs, we tried our best 🤕

.@Cypher_Box is currently only wallet app that does coin selection before amount selection by default. Instant coin labeling on receive is on our roadmap. Many people think this is "advanced UX", not at all. The problem is that bitcoin devs are designing onchain wallets to look like banking wallets, instead of digital cash wallets. Bitcoin educators are then forced to explain coincontrol as an advanced step, not as a basic step for spending bitcoin.

I do not recommend using any hardware wallet without adding layers of security that protect you from the vendor. At least a passphrase and an address derivation check. These companies are not trustworthy.

I've been spending my free time and tokens for the past month Red Teaming many Bitcoin projects sending disclosers and testing. I've seen some dark shit. There are lots of medium-high findings and few critical ones. Most are rookie bugs, very easy to avoid and fix. But human error is unavoidable. I hope people do the same and audit other projects like the one I'm building @Cypher_Box

Ark UX is super fun and easy. Just make sure you buckle your belt and suspenders before you go out cruising 🛥️ - 12 word backup - Auto-foreground-refresh: ON - Scheduled notifications: ON - Emergency exit funded ✅