THE DOSSIER · one of 50

An 839-millisecond price pulse—and 180 deviations

0x9df9ed63122824b2c77e52370bd2b9897df44f98c780f355702945f8aa923285
47,626 transactions in the window · browse every one → · meet The Price-Keeper →

For the frozen window on 2026-07-19, this address repeated one two-call price-update sequence 47,446 times. Its median gap was 839 ms, with a 923 ms p90 gap and only a 2.478-second longest idle. The cadence is consistent with automated oracle-feed maintenance, but the evidence does not establish a person, company, operator, or bot behind the address.

The complete-history brief covers 47,626 eligible transactions from 13:01:48.968 to 23:37:41.663 UTC. Of those, 47,446 transactions—99.62%—asserted update-price. No execution error was recorded anywhere in the history. The remaining 180 transactions, all successful, used a different two-call composition.

What the repeated path did

The repeated pattern, sui/update-price/dual-expiry-write/v1, ran two calls to price_oracle::update_price_with_expiry in the same package revision, 0xf6a6d2492e7b575f3551602652bcd4ff72cc522a42ad5bbfd8c248ed545c4b1b.

The authored explanation describes an access-controlled oracle path that checks updater authority, pair support, expiry bounds, and monotonicity before its insertion or replacement logic proceeds. That explains the mechanism being attempted; it does not prove that an oracle record changed.

At the first observed transaction, 9kpejPxN, the submitted price input was 0.75038526728303 USDC per SUI. At the final transaction, 32ocezVC, it was 0.747400754058395 USDC per SUI. These are call inputs, not proven stored prices, trades, or economic outcomes.

The deviation

All 180 outliers shared one other two-call shape: two calls to price_oracle::halt_price. A representative row, J4MKmnNX, ran at 23:00:54.005 UTC. The calls were structurally identified, but no whole-transaction behavior was asserted for them. The evidence therefore supports “called halt_price twice,” not “the oracle was halted” or “its state changed.”

The detailed ledger is capped at the latest 500 classifiable transactions. Every row in that cap was the ordinary update-price pattern, and the observed gaps ranged from 611 to 965 ms, with a 839 ms median. The busiest minute in that tail contained 76 transactions; across the complete history, the busiest minute was 13:49 UTC, also with 76.

Money and boundaries

The address paid its own gas in all 47,626 historical transactions; none were sponsored. The complete-history gas total was 11.777338288 SUI, or 247,288 MIST per transaction. The same 11.777338288 SUI also appears as a sender holding decrease. Those are two views of the same gas charge and must not be added together.

No USDC movement or other application-value movement is proven for the sender. The submitted price observations therefore do not establish a swap, profit, price correctness, or an external market relationship.

This dossier is bounded to the frozen 2026-07-19 corpus window. The useful exits are the address itself, the repeated pattern sui/update-price/dual-expiry-write/v1, revision 0xf6a6d249…4c4b1b, and the representative digests 9kpejPxN, 32ocezVC, and J4MKmnNX.

Dossiers are AI-authored readings of one address's stored window. They describe observed behavior — never operators, identity, or profit the evidence doesn't establish.

everything this address did → all dossiers the shelf