liquidity that follows demand
Liquiditywhere it'sneeded.Across chains, in seconds.
One reactive contract watches Uniswap v4 pools on Unichain Sepolia and Base Sepolia. When a position falls out of range it re-centres it. When trading shifts to the other chain, it moves liquidity there. No keeper server to run.
median cross-chain move · 5 live moves, p95 4.0 s
Watch. Decide. Act.
three jobs, one contract on reactive, two vaults on the chains
- 1
Watch
The brain subscribes to swap events from the Respool pool on both chains, and to each vault's own events.
- 2
Decide
A swap that leaves a vault's range asks for a re-centre. Recent swap activity three times higher on one chain asks for a move.
- 3
Act
Reactive delivers a callback. The vault re-centres its position, or burns tokens here so the same amount can be minted on the other chain.
Event. Reactive. Callback.
every loop here is a real one, recorded on testnet. pick one and replay it.
- 1Swap on Base Sepoliat + 0.00 sblock clock
- 2Brain reacts on Lasna≈ 1.29 swall clock, scaled
- 3Vault re-centres on Baset + 1.80 sblock clock
Seconds, measured.
median and p95 for the current batch. full-loop time from the chains' own block timestamps.
median · p95 2.4 s · 17 loops
median · p95 3.0 s · 17 loops · whole-second clock, ±1 s
median · p95 4.0 s · 5 loops · small sample
Every loop, one dot each
click a dot to replay it above. the green line is the median, the dashed red line is p95.
All four batches, as a table
| Batch | Recentre Base | Recentre Unichain | Moves | Moves went | Top-ups |
|---|---|---|---|---|---|
| Batch 1 first live run | 1.6 s (p95 3.0, n 16) | 2.0 s (p95 3.0, n 15) | 3.8 s (p95 4.2, n 9) | 9 to Base, 0 to Unichain | 13 |
| Batch 2 after the drain fix | 1.4 s (p95 2.6, n 15) | 2.0 s (p95 3.0, n 15) | 3.6 s (p95 4.6, n 15) | 15 to Base, 0 to Unichain | 2 |
| Batch 3 volume-based demand | 1.4 s (p95 2.0, n 17) | 2.0 s (p95 3.0, n 17) | 3.0 s (p95 4.0, n 11) | 0 to Base, 11 to Unichain | 1 |
| Batch 4 swap-count demand (current) | 1.6 s (p95 2.4, n 17) | 2.0 s (p95 3.0, n 17) | 3.4 s (p95 4.0, n 5) | 2 to Base, 3 to Unichain | 0 |
Unichain Sepolia timestamps are whole seconds, so its numbers carry ±1 s. Five moves is a small sample. There is no Ethereum Sepolia baseline yet. Earlier batches moved liquidity one way only; the rule was reworked, and the table shows that history rather than hiding it.
Where the liquidity is.
read from the chains by your browser, right now. if a node is unreachable it says so, and you can retry.
Base Sepolia
- Loading
- 0000
Unichain Sepolia
- Loading
- 0000
Brain · Reactive Lasna
brain 0x1BF3…4f56 ↗- Loading
- 0000
What it is not.
stated up front, so nobody has to guess
Test tokens
Moved liquidity is Respool's own test tokens, burned on one chain and minted on the other. There is no real bridge and no real money.
Scripted traffic
Every swap in these runs came from my own script, evenly spaced. The move rule has not yet met real trading.
A move takes seconds
A burst of trading shorter than a move is over before the liquidity arrives. That bounds what this can chase.
Clock resolution
Base Sepolia reports milliseconds. Unichain Sepolia and Lasna report whole seconds, so their times are ±1 s.
One-sided after a crossing
When price runs through a position the vault holds one token. A limit position parks it beside the price instead of leaving it idle.
No baseline yet
A run with Ethereum Sepolia as the origin chain, to show what a slow origin costs, is still to do.
On chain, checkable.
the contracts these numbers came from
Run it yourself.
# needs Foundry and Node 18+ $ git clone https://github.com/dev-doji/respool.git $ cd respool $ git submodule update --init $ git -C lib/v4-core submodule update --init lib/solmate $ forge test # 57 tests $ node observer/observe.mjs summary