AquaQoSVerify public proof

PROTOCOL EVIDENCE

Open the receipts.

A short path from the claim to the code, the test and the recorded result.

View onchain proof
SUPPORTED DOMAINone token pair

One token pair.
Up to eight strategies.
Explicit capacity rules.

CAPACITY_GUARD · 0x05Canonical XYCPublic Sepolia receiptsSourcify exact match
EVIDENCE SCOPE22 public txs

AquaQoS adds a custom SwapVM instruction and a restricted maker vault. Official Aqua still performs virtual accounting and ERC-20 settlement.

The comparisons use pinned, fee-free XYC programs and retained receipts. A public Ethereum Sepolia deployment records exact-match verified contracts and 22 transactions. There is no external audit and no mainnet deployment.

72 fixtures · 616 swaps · 12 pushes

Follow the evidence path.

01 / CURATED_EVIDENCEOPENProblem and mechanismWhy independent virtual ledgers can compete for one shared ERC-20 inventory, and where the guard enters the SwapVM path.Open evidence02 / CURATED_EVIDENCEOPENCapacity modelThe admission invariant, allowance floor, protected entitlement and burst semantics.Open evidence03 / CURATED_EVIDENCEOPENMeasured trade-offsConservative Aqua, raw overcommitment and two protection policies across 2, 4 and 8 strategies.Open evidence04 / CURATED_EVIDENCEOPENSecurity model and limitsTrust boundaries, internal review scope, remaining risks and explicit prototype limits.Open evidence05 / CURATED_EVIDENCEOPENPublic Sepolia receiptsExact-match verified contracts and representative protected-capacity transactions.Open evidence06 / CURATED_EVIDENCEOPENSources and licensesPinned official dependencies, retained notices and source material behind the product.Open evidence

What the prototype does—and where it stops.

KNOWN LIMITS4 items

  1. Smaller guarantees leave more inventory for burst. Higher fill volume at 50% protection is not an equal-protection efficiency claim.

  2. At eight strategies, low-contention guarded swaps measured 210,099 median gas versus 113,715 raw. Guard rejection can cost more than failed settlement.

  3. Transaction-scoped reservations can reject otherwise safe sequential fills in the same outer transaction.

  4. Fee-on-transfer, rebasing and malicious tokens are outside the supported domain. These tests do not establish general solvency or financial safety.

SourcesSecurity modelBenchmark methodologySources and licenses