Satsuma
S

@satsuma/experiments

PublicThe system
Clone

Governed runs of Satsuma trading on tangerine over ISMP. Results are published as articles when a run completes.

Lab runs, not the live system. Each run uses lab chains that are reset afterwards, with synthetic users unless a run says otherwise. Nothing here happened on the Satsuma chain.
RunsLab run · 10 November 2026, 20:48 UTC

The test bench

Can a run be started from a saved chain state and its model calls answered from a recording, so that the same result comes back without calling the model again?

1/5 passed4 inconclusive · 1 h 37 min

Setup

Chains
Satsuma lab chain
  • RoleHome
  • KeptLab, not retained
tangerine lab chain
  • RoleProving
  • KeptLab, not retained
Users
  • UsersSynthetic
Model
Claude Opus
  • ModeReplayed
Operator seat
  • SeatBy delegate
Unattended
Yes: nobody was watching to step in
Time
1 h 37 min · ended 22:25 UTC

Checklist 5 items

Every item, unless its row says otherwise
  • ChainLab
  • UsersSynthetic
  • DecisionsReplayed

Rolling back a bad change

0/4 passed
InconclusiveAn upgrade that made things worse is rolled back to the version before it6 of 7 attempts
InconclusiveThe rollback is triggered by the trading account's own records, not by an alarm on the contract6 of 7 attempts
InconclusiveThe rolled-back change is recorded as an incident6 of 7 attempts
InconclusiveTrading is healthy again after the rollback6 of 7 attempts

Replayable test runs

1/1 passed
PassBuilt products are installed from a shelf of earlier builds onto a fresh chain and adopted through the normal path, with no new build1 of 1 attempts

What this proves

  • Whether recorded model answers replay to the same verdicts, and whether built products can be installed from earlier builds instead of building them again.

What it does not

  • The model's judgment on anything outside what was recorded.
  • Anything about the live system.

What broke, what was fixed

The record of this run does not say whether a person stepped in, so it does not count as a run without repairs.

Cost in tokens

Not recorded: this run has no chain-derived scorecard.

Archived evidence

The lab chains this run used were reset afterwards and are not retained, so nothing here links to an explorer. The record below is the archived copy.

Scorecard
None archived.