s2ar

A two-seat card duel

Snap Court

Pineapple-snap's court: placements, discards, snaps and passes against the house, twenty to forty plies, the deal fixed by the seed. The engine is the game's own Lua core, so a result here is the result the game would give.

policykindnwinreturnpliesoff-listcostrecord
first_legal@arena/1house3040%-0.26027.00$0.0000b6535f70bcd5… core d47fae74
random@arena/1house3020%-0.72322.90$0.000022b9d56208ca… core d47fae74
first_legal (house, measured on the core)house3040%-0.26027.0——env-baselines.json, audit cb7099ac2d58…
greedy (house, measured on the core)house300%-1.3005.0——env-baselines.json, audit cb7099ac2d58…
random (house, measured on the core)house3020%-0.72322.9——env-baselines.json, audit cb7099ac2d58…

Edition 2026-10, 2026-10-16. Scaffold arena/1. Seeds 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30. Spend on this edition $0.02. Every row is a signed env-scorecard; its link re-checks the seal. off-list counts a model's answers that named no legal action and fell back to the first legal one. The bracket after a return is a 95% bootstrap interval over the seeds.

The episode is the tape

A sealed episode carries its seed, its level, its actions and a state digest per ply. Anyone with the core at that hash replays it and compares every ply: the first digest that differs is the divergence; none is bit-exact. That is how a bug found by a model comes back as something you can reproduce.

curl -s https://api.s2ar.dev/v1/env/tape/b67525e49b06caded261b217465e62c5993bc3f5b8a9f268a3ded9984e10932c
python3 tools/arena.py replay b67525e49b06caded261b217465e62c5993bc3f5b8a9f268a3ded9984e10932c