When the books disagree,
the line blocks itself.
One rule · three committed runs · route A, testnet
Same contract, three outcomes
Forged mintbu-015
Lock arrives latelf-009
In orderls-015
Two boxes, no interlock
Each chain keeps half the book
Nomad: about $190M lost to copycat replays in roughly 2.5 hours.
Base sees
Locked · collateral in
Unichain sees
Minted · tokens out
One instrument, both ends
One contract reads both
A Reactive Omni contract on Lasna keeps both books and pauses both gates when they disagree.

The interlock, step by step
Pause first, match later
Status page · last non-control attack run
Four timestamps, four hashes

- 1mint mined
- 2react() seen
- 3gate B Paused
- 4gate U Paused
100 attack runs · final contract
Both gates shut, every run
Lasna react() seen: 3.873 s (p50). In-order legit transfers paused: 0 of 80.
Control · 5 runs per arm
Same attack script, with and without
Unguarded
Ledgerlock
Fixed 6 copycats vs up to 30: illustration, not rate.
Honest limits, abridged · README
What it does not do
- 02Mints before the pause are lost, the first always.
- 05Route A is our fixture; route B our own warp route.
- 06Reorgs: Test Lib simulation only.
The ask
Judge us on the seconds
repo github.com/edycutjong/ledgerlock
live edycutjong.github.io/ledgerlock
judge edycutjong.github.io/ledgerlock/judge
release v0.5.5
testnet Lasna · Base Sepolia · Unichain Sepolia
Every measured number comes from a committed row.