The XRP Ledger has a bug-fix bundle queued for mainnet activation on 11 September. It hardens Single Asset Vaults, the Lending Protocol, automated market makers, the permissioned DEX, Checks and pseudo-accounts. The awkward part is what sits underneath that list: the Lending Protocol it patches is not live, and the amendments that would make it live are nowhere near the votes they need.
The amendment is fixCleanup3_3_0. It shipped with xrpld version 3.3.0, entered validator voting on 6 August, and crossed the 80% threshold on 28 August with 82.86% support on 29 Yes votes. Crossing that line opens a 14-day activation window, and XRPScan data puts the earliest mainnet activation at 11 September 2026.
“Earliest” is carrying weight there. The 80% floor is continuous rather than a one-time gate: if support drops to 80% or below at any point inside the window, the amendment is rejected and the fourteen days start over. Ripple engineer Hussein Zangana, flagging the milestone publicly, told node operators to upgrade in time so their nodes keep operating once it activates — nodes left on older software stop following consensus at the switch.
What is actually in the bundle
This is maintenance, not a headline feature. fixCleanup3_3_0 unifies freeze and deep-freeze checks for transfers to and from pseudo-accounts across vault, AMM and loan-broker transactions. It makes CheckCash and CheckCancel reject an all-zero CheckID at preflight with temMALFORMED rather than failing later in processing. It stops hybrid offers vanishing from the open order book when the account that placed them loses access to a permissioned domain, and folds AMM liquidity into quality estimates for permissioned DEX order books.
There is also a cluster of arithmetic hardening: AMMWithdraw returns tecAMM_FAILED instead of dividing by zero at the one EPrice value where the denominator collapses, and AMMDeposit, AMMWithdraw and AMMClawback gain precision-loss checks when fixAMMv1_3 is enabled. Two invariants get tightened so an AMM can only be removed through a legitimate path, and so deleting a ledger entry backed by a pseudo-account deletes that pseudo-account too.
None of that is glamorous. All of it is the kind of edge case that turns into a post-mortem when real money is sitting in the vault.
The sequencing is the story
Read the two votes side by side and a pattern appears. The bug fixes for on-chain credit cleared 80% comfortably. The credit itself has not: the native lending amendments — XLS-65 and XLS-66, which would enable fixed-term, pooled, uncollateralised loans — have been tracking near 40% support, roughly half what activation requires. We covered that stall on Monday, and nothing in this week's vote changes it.
That is validators doing something fairly sensible: accepting the plumbing while withholding the product. Freeze semantics, invariant checks and precision handling are hard to argue against. A new uncollateralised credit primitive on a payments chain is a different conversation, and one a meaningful bloc of validators is clearly not finished having.
Our take: The interesting number is not 82.86%. It is the gap between that and the ~40% behind the lending amendments themselves. Institutions watching XRPL as tokenised-credit infrastructure keep reading amendment votes as adoption signals; this pair says the opposite. The ledger is being made safe for a feature its own validators have not agreed to turn on. That is a healthier order of operations than the reverse — but anyone modelling XRPL lending as a 2026 revenue line is modelling a vote that has not happened.
What to watch
- 11 September. The floor, not a promise. Watch whether support holds above 80% through the window or resets the clock.
- Node upgrade rates. Operators still on pre-3.3.0 builds at activation drop off consensus. A messy switchover is the visible failure mode.
- XLS-65 and XLS-66 movement. If lending support starts climbing off ~40% in the weeks after this activates, the fix bundle was the blocker. If it stays flat, the objection is to the product.
- Whether anything reaches production. Vaults and permissioned DEX features only matter once real balances use them. Amendment votes are not usage.
Price context, for what it is worth: XRP has spent the week trading in the low $1.30s to high $1.40s after August's run stalled on profit-taking. None of that is connected to a maintenance amendment, and it would be a mistake to pretend otherwise. Governance moves on its own clock.