TRUC Transactions (v3)

TRUC transactions — Topologically Restricted Until Confirmation, commonly called "v3" — are an opt-in mempool policy signalled by setting a transaction's nVersion field to 3. Specified in BIP 431 and made standard policy in Bitcoin Core 28.0, TRUC restricts how an unconfirmed transaction can be chained with other unconfirmed transactions, so that fee-bumping stays predictable and resistant to pinning attacks. It is a policy rule, not a consensus rule: nodes agree to relay and prioritize v3 transactions under these constraints, but nothing about block validity changes.

The topology restrictions

A TRUC transaction accepts a deliberately tiny family. It may have at most one unconfirmed ancestor and at most one unconfirmed descendant — a straight line of two transactions, never a tree. The unconfirmed child is additionally capped in size (roughly 1,000 virtual bytes) so it cannot be bloated, and v3 transactions may only have v3 unconfirmed parents, keeping the regime closed. Compare that with default policy, where a single transaction can drag along a deep graph of unconfirmed descendants: under those rules, Replace-by-Fee gets expensive precisely because a replacement must outbid everything hanging off the original. By keeping the cluster tiny and fully understood, a node can always evaluate a replacement at the package level without an adversary being able to weaponize the descendant limit.

Why second layers needed it

Contracting protocols such as the Lightning Network depend on broadcasting a presigned commitment transaction — signed possibly weeks earlier, at a fee that is stale by the time it matters — and then attaching fees at broadcast time via a child (CPFP). Under older mempool rules an attacker who could add their own spend to the commitment could "pin" it: append a large, low-feerate descendant that makes replacing or bumping the package uneconomical, stalling the honest party's transaction exactly when channel-security timelocks are ticking. TRUC closes that door structurally — there is simply no room in the topology for the junk child. It also enables a striking property: a v3 parent may pay zero fee (below the minimum relay feerate) as long as a fee-paying child carries the package over the line, which is the basis of ephemeral anchors and lets protocol designers stop guessing future feerates at signing time altogether.

Limits and context

The protection is per-mempool policy, so its strength scales with adoption across relaying nodes; a miner running permissive custom policy is not bound by it, though incentive-compatible package evaluation means the honest configuration is also the profitable one. TRUC is one piece of the broader package-relay effort in Bitcoin Core — alongside package RBF and one-parent-one-child relay — that collectively make lightweight fee-bumping dependable for time-sensitive protocols. From a decentralization angle this matters more than it looks: reliable fee-bumping is what lets self-hosted Lightning nodes on modest hardware defend their channels without overpaying insurance fees up front.

TRUC ships alongside a companion mechanism worth knowing by name: ephemeral anchors, standardized around a tiny "pay-to-anchor" (P2A) output that anyone can spend. A protocol transaction carries a zero-value anchor output, pays no fee itself, and whoever needs it confirmed attaches a fee-paying child spending the anchor — with TRUC's topology guaranteeing the package stays small and replaceable. This division of labour (consensus-valid transaction now, fee decision later, by whichever party cares) is quietly becoming the standard design idiom for everything from Lightning commitments to more exotic contracting protocols, and it only works because v3's restrictions make the child's behaviour predictable. Expect the version-3 marker to become steadily more common in mempools as Lightning implementations and other contract protocols migrate their transaction formats onto it.

For the attack class TRUC neutralizes, see transaction pinning; for the replacement mechanics it makes dependable, see Replace-by-Fee.