GameBench OPEN BENCHMARK
reproduce · level 2 · Benchmark 0.2.0

Reproduce OhSteem

复刻 OhSteem · reproduce.ohsteem.v1

1.0.0task version
12atomic tests
60m agent budget
0published runs
Generated outputs

Every seed stays visible.

Official output defaults to seed 104729. The gallery never chooses the highest-scoring attempt, and failures remain in the record.

No published output for this release. The frozen task contract remains available for audit.
Test contract

What this game measures.

buildmechanicsfeelvisualstability
Atomic testCategoryPointsCase
build build 5 build
cell-cycle mechanics 10 cell-cycle
no-three mechanics 10 no-three
balanced-colors mechanics 10 balanced-colors
unique-lines mechanics 10 unique-lines
completion mechanics 5 completion
input-feedback feel 10 input-feedback
size-reset feel 10 size-reset
visual-home visual 10 visual-home
visual-board visual 10 visual-board
deterministic-puzzle stability 5 deterministic-puzzle
locked-cell stability 5 locked-cell

Runtime contract

The game is evaluated at 1280 × 720 with Node 22, deterministic seed reset, controlled time, native input, and schema-validated snapshots.

Reference: OhSteem at commit 61bb4e5c3b9d2e5704b805edb994de66214ed1b0, license MIT.

Frozen identity

Task content
sha256:7bc9064a12cb2c022401ff2706d6c1cb5080e4d3e2af32379aff5bb285bb584c

Benchmark release
v0.2.0 release contract