At 13:25 UTC, this Sui minute moved from routine order maintenance into a burst of swap attempts. It contained 1,654 transactions, of which 296 failed—17.9%, more than twice the frozen corpus rate of 8.36%. The failures were not one clean outage: 60 had no recorded abort site, 52 stopped while returning a flash-loan quote, and 54 stopped in two closely related pool-swap functions.
The minute was almost continuous activity. Its median gap was 0 milliseconds, its mean gap 36 milliseconds, and its longest idle period only 624 milliseconds. The first 134 transactions were led by limit-order placement (25) and cancellation (24). In the next five-second bucket, swap routes became the leading named behavior with 47 of 183 transactions, while 41 more were understood only as linked call fragments.
A concentrated opening act
One sender, 0xcde6db…6c0e04, accounted for 332 of the 1,654 transactions—20.1% of the minute. Its activity was unusually consistent: 221 cancellation attempts and 111 limit-order placement attempts, all using DeepBook-oriented patterns. Only one of those 332 transactions failed. The repetition is consistent with highly automated execution, but the address itself does not establish a person, company, or operator.
The transaction details show what these calls meant—and what they did not prove. A cancellation such as 2cuMbK72 passed an owner proof, identified an order, and described its resting price and size. That supports an order-cancellation action; it does not prove that the order had traded. Similarly, placement calls recorded resting prices and submitted sizes without proving that an order entered or remained on the book.
A previously unobserved revision enters the failure burst
The store places 12 exact revision minima in this minute, more than 23 minutes after observation began. One is revision 0x0489a4…ead065: six transactions used it in the minute, four of them failed. Its first stored example, 3i1YQBpr at 13:25:09.189 UTC, withdrew vault-held USDC, passed it through an adapter touching WAL and USDC, opened a flash swap, prepared repayment, closed the flash position, and deposited the result back into the vault. The route submitted 10 USDC and guarded on a 10 USDC minimum output. This is a genuine entry into the corpus's observed repertoire, not proof that the package was deployed then. The stored meaning is precise about the sequence but does not establish that an exchange occurred.
The next revision-linked attempts show why the minute’s failure rate spiked. E5r5E3Pm, only 215 milliseconds later, reached a slippage verification call guarding on 48.191833 USDC and aborted there. Application effects were rolled back; the transaction proves an attempted route and its failure boundary, not a completed conversion. Earlier, HtNXdfd2 attempted a similar vault-funded route with 3,091.214732585 WAL, reached the pool swap, and stopped at pool::swap_in_pool.
Taken together, these records suggest a route family being exercised across several variants, with failures occurring at pool, repayment, slippage, and balance-validation boundaries. They do not identify the cause as a single protocol-wide fault, nor do they prove that any route completed an economic exchange.
What moved
Sender-boundary movement evidence recorded net positive SUI of 59,758.633533162 and USDC of 1,433.22913, alongside net negative WAL of 315,761.444288528 and positive DEEP of 4,197,923.49347. Separately, gas movement totaled −1.513777402 SUI. These are per-asset totals from the minute’s movement evidence; they are not prices, profit, or a cross-asset result, and the gas figure must not be added to the application-value figures.
The strongest reading of this minute is therefore a change in the observed activity regime: a fast, repetitive DeepBook maintenance stream was joined by a previously unobserved, more complicated swap-route revision whose attempts repeatedly reached meaningful safeguards and then failed. The chain records the sequence clearly. It does not establish when the package was deployed, whether the underlying strategy was profitable, who controlled the addresses, or whether any attempted swap settled.
Useful exits are 3i1YQBpr for the committed route sequence, E5r5E3Pm for the slippage failure, HtNXdfd2 for the pool failure, and revision 0x0489a4…ead065 for the route machinery first stored in this minute.