Move the round strip out from under the opponent hand; let undo cross turns

Two fixes to the last change:

- The round strip sat top-center, exactly where the opponent's hand is, so its
  cards covered it. Moved it under the rival plaque on the left, clear of the hand.

- Undo/redo was capped at the current turn. When the setting is on it now walks the
  whole timeline -- back through committed moves, the rival's replies, and earlier
  rounds. The rival stays suspended whenever there are moves ahead to redo, so
  reviewing never makes it replay the move being looked at. Menu label updated to
  "전체 되감기".

Verified in a browser: the strip no longer overlaps the opponent hand, and undo
from mid-game walks back eight steps across a rival turn to the opening.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XBQKgvBbxbheiTF1AVy1Sh

.gitignore

@@ -15,6 +15,9 @@ src/**/*.c
 # Local session / lock files
 .compute.lock
 .claude/
+# ...except checked-in project skills, which are documentation for the next session.
+!web/.claude/
+!web/.claude/**
 
 # Rust build output
 target/
@@ -29,10 +32,13 @@ tools/julia/
 .ruff_cache
 
 # Web dependencies and locally exported models. The verified public browser
-# policy below is the one exception: it is deliberately served as a static asset.
+# policies below are the exception: they are deliberately served as static assets,
+# and the deploy builds straight from the repo, so an ignored model ships as a 404.
 web/node_modules/
 web/*.tsbuildinfo
 web/public/models/*.onnx
 web/public/models/*.json
 !web/public/models/jax-ppo.onnx
 !web/public/models/jax-ppo.json
+!web/public/models/borealis.onnx
+!web/public/models/borealis.json

data/models.json

@@ -0,0 +1,43 @@
+{
+  "_scheme": {
+    "naming": "Astronomical names, alphabetically ordered. The first letter is the generation; a new letter means the observation space broke, not that the model got better.",
+    "rule": "A codename never encodes quality. The record this replaces stored 'FINAL PPO', which stops meaning anything the moment there is a second final model.",
+    "identity": "The hash is the truth -- it is what actually played. The codename is for humans, and it is assigned here, not derived. Records should carry the hash; look the codename up.",
+    "next": "cygnus, deneb, ..."
+  },
+
+  "models": {
+    "altair": {
+      "hash": "e8241e305c01",
+      "hash_kind": "sha256 of web/public/models/jax-ppo.onnx",
+      "generation": "a",
+      "game": "single round",
+      "observation_size": 454,
+      "hidden_size": 512,
+      "num_layers": 3,
+      "trained": "self-play league, 122.6M learner actions",
+      "source": "/mnt/2tbhdd/coolrl-lost-cities-artifacts/league/2026-07-05_052325_jax-ppo-league-v1/latest",
+      "deployed": "web/public/models/jax-ppo.onnx",
+      "displayed_as": "WASM · FINAL PPO",
+      "note": "Every game in data/human-play/game-records.jsonl (format v1, 111 games, 2026-07-14) was played against this model. The v1 schema has no model field -- it stores the on-screen label -- so this line is the record of what they played."
+    },
+
+    "borealis": {
+      "hash": "4ae613b010ca",
+      "hash_kind": "sha256 over the orbax checkpoint files (no ONNX export yet)",
+      "generation": "b",
+      "game": "three-round match (classic rules)",
+      "observation_size": 501,
+      "critic_observation_size": 681,
+      "hidden_size": 512,
+      "num_layers": 3,
+      "trained": "self-play, 131.1M learner actions, linear total-score reward, both seats, privileged critic",
+      "source": "runs/jax-ppo-match/2026-07-15_031529_match-scaled/latest",
+      "deployed": null,
+      "results": {
+        "vs_altair_3round": "0.6094 win rate [0.599, 0.620], +22.0 points, 8192 duplicate matches",
+        "exploitability": "a from-scratch exploiter funded to 131M reaches 0.4657 against it, and 0.6295 against altair -- lower bound, neither exploiter had plateaued"
+      }
+    }
+  }
+}

docs/plans/lost-cities-classic-3round.md

@@ -581,3 +581,37 @@ Fable은 이것을 "가장 큰 누락 아이디어"로 꼽았다. 실측은 **
 **흥미로운 점:** 우리는 **순수 셀프플레이**이고 league는 **착취자 구조를 학습에 넣은** 런인데,
 그런데도 우리가 덜 착취당한다. 착취자 구조가 exploitability를 낮춰줄 것이라는 기대가
 이 게임에서는 확인되지 않았다.
+
+### 착취자에 제대로 자금을 대고 재측정 (131M, 목표와 동일 예산)
+
+앞선 32.5M 측정은 **공격자가 목표보다 4배 부족**했다. 예산을 목표와 맞춰 다시 쟀다.
+실측 확인: 착취자 learner 액션 **130.4M** (우리 상대) / **131.1M** (league 상대).
+
+| 얼려놓은 정책 | 착취자 승률 | 95% CI | 착취자 평균 마진 | 이전(32.5M) |
+|---|---|---|---|---|
+| **우리 매치 스택** (131.1M 학습) | **0.4657** | [0.455, 0.477] | **−5.5점** | 0.2278 |
+| league (122.6M 학습) | **0.6295** | [0.619, 0.640] | **+28.8점** | 0.3213 |
+
+**CI가 전혀 겹치지 않고, 격차가 오히려 벌어졌다 (0.094 → 0.164).**
+
+**질적으로 선을 넘었다:** 제대로 자금 댄 전담 공격자는 **league를 아예 이긴다**(0.63, +28.8점).
+반면 **우리는 여전히 못 뚫는다**(0.47, −5.5점).
+
+#### 우리에게 불리한 단서 (반드시 함께 읽을 것)
+
+- **둘 다 1000 업데이트 끝까지 정체 없이 오르고 있었다. 이 수치도 여전히 하한선이다.**
+- **우리 쪽 착취자의 말단 기울기가 더 가파르다** (+0.031 vs +0.016 / 100 업데이트).
+  league 착취자 곡선은 꺾이기 시작했는데 우리 쪽은 아직 갈 길이 남았다.
+  **공격자 예산을 크게 더 키우면 격차가 좁혀질 수 있고, 뒤집힐 가능성도 배제 못 한다.**
+
+#### 구조적 교란 (순수한 "학습법 A vs B"가 아니다)
+
+1. **league는 단판 정책이다.** carry도 매치 점수도 모른다. 더 털리는 것의 일부는
+   **학습 방법이 아니라 구조적 맹점** 때문일 수 있다.
+2. **둘 다 greedy(argmax)로 뒀다.** 결정론적 정책은 **정의상 최대로 착취당한다.**
+   양쪽을 똑같이 대우했으니 비교는 공정하나, **절대값은 부풀려져 있다.**
+
+#### 살아남는 주장 / 못 하는 주장
+
+- ✅ **"동일 예산의 전담 공격자 앞에서 league는 뚫리고 우리는 안 뚫린다."**
+- ❌ "우리 정책은 착취 불가능하다." — 하한선일 뿐이고, 우리 쪽 곡선은 아직 오르는 중이다.

scripts/altair_vs_borealis.py

@@ -0,0 +1,201 @@
+"""altair (single-round, gen a) vs borealis (3-round match, gen b).
+
+Full 2x2: {single deal, 3-round match} x {win rate + Wilson, mean margin + CI}.
+
+Both policies act on the SAME MatchState but from their own view:
+- borealis reads the full MatchState (match_observation + privileged critic).
+- altair reads only the round in play as a single-round State (observation).
+
+We play whole matches (duplicated: every deal-triple from both seats). From the
+same runs we harvest two scoring conventions:
+- single deal   = the round-0 board score, snapshotted the ply round 0 rolls over
+                  (carry=0, round_idx=0 there, so it is an honest standalone deal).
+- 3-round match = match_score(final), the summed total.
+
+Margins are borealis-minus-altair. Self-play (borealis vs borealis) is run as a
+harness check: mirrored identical policies must give win rate 0.5 / margin 0.0.
+"""
+
+from __future__ import annotations
+
+import json
+import math
+from pathlib import Path
+
+import jax
+import jax.numpy as jnp
+import numpy as np
+
+from lost_cities_jax.match import (
+    MatchState,
+    match_legal_action_mask,
+    match_reset_from,
+    match_score,
+    match_step,
+)
+from lost_cities_jax.match_eval import MATCH_SCAN_STEPS, _wilson, match_bank
+from lost_cities_jax.match_obs import match_critic_observation, match_observation
+from lost_cities_jax.match_ppo import Ablation, MatchActorCritic, create_match_train_state
+from lost_cities_jax.obs import observation
+from lost_cities_jax.ppo import (
+    ActorCritic,
+    create_train_state,
+    load_config,
+    mask_logits,
+    restore_checkpoint,
+)
+
+ALTAIR_CKPT = Path(
+    "/mnt/2tbhdd/coolrl-lost-cities-artifacts/league/2026-07-05_052325_jax-ppo-league-v1/latest"
+)
+BOREALIS_CKPT = Path("runs/jax-ppo-match/2026-07-15_031529_match-scaled/latest")
+MATCHES = 4096  # -> 8192 duplicate games per cell
+SEED = 20260715
+OUT = Path("runs/jax-ppo-match/altair_vs_borealis.json")
+
+borealis_cfg = load_config("configs/jax_ppo/match-selfplay.yaml")
+altair_cfg = load_config("configs/jax_ppo/balanced.yaml")
+
+ABL = Ablation()  # privileged_critic=True, as borealis was trained
+borealis_st = restore_checkpoint(
+    BOREALIS_CKPT, create_match_train_state(borealis_cfg, jax.random.PRNGKey(0), ABL)
+)
+altair_st = restore_checkpoint(ALTAIR_CKPT, create_train_state(altair_cfg, jax.random.PRNGKey(0)))
+
+borealis_model = MatchActorCritic(borealis_cfg.network.hidden_size, borealis_cfg.network.num_layers)
+altair_model = ActorCritic(altair_cfg.network.hidden_size, altair_cfg.network.num_layers)
+
+
+def _borealis_action(env: MatchState, to_move, mask):
+    obs = jax.vmap(match_observation)(env, to_move)
+    crit = jax.vmap(match_critic_observation)(env, to_move)
+    logits, _ = borealis_model.apply(borealis_st.params, obs, crit)
+    return jnp.argmax(mask_logits(logits, mask), axis=-1).astype(jnp.int32)
+
+
+def _altair_action(env: MatchState, to_move, mask):
+    obs = jax.vmap(observation)(env.round, to_move)
+    logits, _ = altair_model.apply(altair_st.params, obs)
+    return jnp.argmax(mask_logits(logits, mask), axis=-1).astype(jnp.int32)
+
+
+def _make_run(action_seat0, action_seat1):
+    """Build a jitted full-match runner.
+
+    ``action_seat0`` is the policy that plays when ``to_move == borealis_seat``
+    (i.e. borealis); ``action_seat1`` is the other policy (altair). Selection is
+    by ``borealis_seat`` so mirrored orientation is a pure seat relabel.
+    """
+
+    @jax.jit
+    def run(env: MatchState, borealis_seat):
+        def body(carry, _):
+            env, r0_snap = carry
+            to_move = env.round.to_move.astype(jnp.int32)
+            mask = jax.vmap(match_legal_action_mask)(env)
+
+            a0 = action_seat0(env, to_move, mask)
+            a1 = action_seat1(env, to_move, mask)
+            action = jnp.where(to_move == borealis_seat, a0, a1)
+
+            was_r0 = env.round_idx == 0
+            nxt, _, _ = jax.vmap(match_step, in_axes=(0, 0))(env, action)
+            just_finished_r0 = was_r0 & (nxt.round_idx == 1)
+            # nxt.carry == round-0 board score exactly on the roll-over ply.
+            r0_snap = jnp.where(just_finished_r0[:, None], nxt.carry, r0_snap)
+            return (nxt, r0_snap), None
+
+        n = env.done.shape[0]
+        r0_snap = jnp.zeros((n, 2), dtype=jnp.int32)
+        (env, r0_snap), _ = jax.lax.scan(body, (env, r0_snap), xs=None, length=MATCH_SCAN_STEPS)
+        return env, r0_snap
+
+    return run
+
+
+def _summ(margin: np.ndarray) -> dict:
+    """margin = borealis - altair, per duplicate game. Positive = borealis wins."""
+    n = int(margin.size)
+    b_wins = int((margin > 0).sum())
+    a_wins = int((margin < 0).sum())
+    ties = int((margin == 0).sum())
+    lo, hi = _wilson(float(b_wins), float(n))
+    std = float(margin.std(ddof=1))
+    sem = std / math.sqrt(n)
+    return {
+        "n_duplicate_games": n,
+        "borealis_wins": b_wins,
+        "altair_wins": a_wins,
+        "ties": ties,
+        "borealis_win_rate": b_wins / n,
+        "wilson_95": [lo, hi],
+        "mean_margin_borealis_minus_altair": float(margin.mean()),
+        "margin_std": std,
+        "margin_sem": sem,
+        "margin_95ci": [float(margin.mean() - 1.96 * sem), float(margin.mean() + 1.96 * sem)],
+    }
+
+
+def _play(run, decks, coins, borealis_first: bool):
+    """Duplicate play; returns (single_deal_margins, match_margins).
+
+    ``borealis_first`` picks which policy is action_seat0 in the runner. When the
+    two policies are identical (self-play) this must yield perfectly antisymmetric
+    margins -> win rate 0.5, margin 0.
+    """
+    single, match = [], []
+    for seat in (0, 1):
+        env = jax.vmap(match_reset_from)(decks, coins)
+        b_seat = jnp.full((MATCHES,), seat, dtype=jnp.int32)
+        final, r0 = run(env, b_seat)
+        r0 = np.asarray(r0)
+        totals = np.asarray(jax.vmap(match_score)(final))
+        # borealis is at index ``seat``.
+        single.append(r0[:, seat] - r0[:, 1 - seat])
+        match.append(totals[:, seat] - totals[:, 1 - seat])
+    return np.concatenate(single), np.concatenate(match)
+
+
+def main():
+    decks, coins = match_bank(SEED, MATCHES)
+
+    # --- Harness check: borealis vs borealis (both seats borealis) ---
+    run_self = _make_run(_borealis_action, _borealis_action)
+    self_single, self_match = _play(run_self, decks, coins, True)
+    self_check = {
+        "single_deal": _summ(self_single),
+        "three_round_match": _summ(self_match),
+    }
+
+    # --- Real comparison: altair vs borealis ---
+    # seat0-slot = borealis (selected when to_move == borealis_seat), seat1 = altair
+    run_av = _make_run(_borealis_action, _altair_action)
+    av_single, av_match = _play(run_av, decks, coins, True)
+    comparison = {
+        "single_deal": _summ(av_single),
+        "three_round_match": _summ(av_match),
+    }
+
+    result = {
+        "meta": {
+            "altair_ckpt": str(ALTAIR_CKPT),
+            "borealis_ckpt": str(BOREALIS_CKPT),
+            "matches_per_orientation": MATCHES,
+            "duplicate_games_per_cell": 2 * MATCHES,
+            "match_scan_steps": MATCH_SCAN_STEPS,
+            "seed": SEED,
+            "privileged_critic": ABL.privileged_critic,
+            "policies": "greedy argmax (deterministic given the deal)",
+            "margin_sign": "borealis total minus altair total",
+        },
+        "harness_check_borealis_vs_borealis": self_check,
+        "altair_vs_borealis": comparison,
+    }
+
+    OUT.parent.mkdir(parents=True, exist_ok=True)
+    OUT.write_text(json.dumps(result, indent=2))
+    print(json.dumps(result, indent=2))
+
+
+if __name__ == "__main__":
+    main()

scripts/export_match_onnx.py

@@ -0,0 +1,138 @@
+#!/usr/bin/env python3
+"""Export a match policy's actor trunk to ONNX for the browser.
+
+Only the actor ships. The critic exists to grade moves during training and never
+plays, so its trunk -- and the privileged view of the opponent's hand and the deck
+that feeds it -- is dropped here rather than shipped and then not used. That also
+means the exported graph physically cannot leak hidden state, which is a stronger
+guarantee than promising not to call it.
+
+MatchActorCritic lays the actor out as Dense_0..Dense_{num_layers} exactly as the
+single-round model does, so the graph construction is the same; only the input
+width and the checkpoint loader differ.
+"""
+
+from __future__ import annotations
+
+import argparse
+import hashlib
+import json
+from pathlib import Path
+
+import jax
+import jax.numpy as jnp
+import numpy as np
+
+from lost_cities_jax.match_obs import MATCH_CRITIC_OBS_DIM, MATCH_OBS_DIM
+from lost_cities_jax.match_ppo import Ablation, MatchActorCritic, create_match_train_state
+from lost_cities_jax.ppo import load_config, restore_checkpoint
+from lost_cities_jax.types import N_ACTIONS
+
+
+def build_argparser() -> argparse.ArgumentParser:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("--checkpoint", type=Path, required=True)
+    parser.add_argument("--config", type=Path, default=Path("configs/jax_ppo/match-selfplay.yaml"))
+    parser.add_argument("--output", type=Path, required=True)
+    parser.add_argument("--codename", required=True, help="see data/models.json")
+    return parser
+
+
+def export_model(checkpoint: Path, config: Path, output: Path, codename: str) -> None:
+    try:
+        import onnx
+        from onnx import TensorProto, helper, numpy_helper
+        from onnx.reference import ReferenceEvaluator
+    except ImportError as exc:
+        raise SystemExit("onnx is required; run with `uv run --with onnx ...`") from exc
+
+    cfg = load_config(config)
+    state = restore_checkpoint(
+        checkpoint, create_match_train_state(cfg, jax.random.PRNGKey(0), Ablation())
+    )
+    params = state.params
+    dense = params["params"]
+
+    nodes = []
+    initializers = []
+    previous = "obs"
+    for index in range(cfg.network.num_layers):
+        layer = dense[f"Dense_{index}"]
+        weight, bias = f"dense_{index}.weight", f"dense_{index}.bias"
+        initializers.extend(
+            [
+                numpy_helper.from_array(np.asarray(layer["kernel"], dtype=np.float32), weight),
+                numpy_helper.from_array(np.asarray(layer["bias"], dtype=np.float32), bias),
+            ]
+        )
+        nodes.append(helper.make_node("Gemm", [previous, weight, bias], [f"dense_{index}.linear"]))
+        nodes.append(helper.make_node("Relu", [f"dense_{index}.linear"], [f"dense_{index}.relu"]))
+        previous = f"dense_{index}.relu"
+
+    actor = dense[f"Dense_{cfg.network.num_layers}"]
+    initializers.extend(
+        [
+            numpy_helper.from_array(np.asarray(actor["kernel"], dtype=np.float32), "actor.weight"),
+            numpy_helper.from_array(np.asarray(actor["bias"], dtype=np.float32), "actor.bias"),
+        ]
+    )
+    nodes.append(helper.make_node("Gemm", [previous, "actor.weight", "actor.bias"], ["logits"]))
+
+    graph = helper.make_graph(
+        nodes,
+        f"coolrl-lost-cities-match-actor-{codename}",
+        [helper.make_tensor_value_info("obs", TensorProto.FLOAT, [None, MATCH_OBS_DIM])],
+        [helper.make_tensor_value_info("logits", TensorProto.FLOAT, [None, N_ACTIONS])],
+        initializer=initializers,
+    )
+    model = helper.make_model(
+        graph, producer_name="coolrl-lost-cities", opset_imports=[helper.make_opsetid("", 17)]
+    )
+    model.ir_version = 8
+    onnx.checker.check_model(model)
+
+    # The exported graph has to agree with the trained one, not merely load.
+    rng = np.random.default_rng(20260715)
+    sample = rng.normal(size=(8, MATCH_OBS_DIM)).astype(np.float32)
+    critic_stub = jnp.zeros((8, MATCH_CRITIC_OBS_DIM), dtype=jnp.float32)
+    flax_model = MatchActorCritic(cfg.network.hidden_size, cfg.network.num_layers)
+    expected, _ = flax_model.apply(params, jnp.asarray(sample), critic_stub)
+    actual = ReferenceEvaluator(model).run(None, {"obs": sample})[0]
+    np.testing.assert_allclose(actual, np.asarray(expected), rtol=2e-5, atol=2e-5)
+    np.testing.assert_array_equal(
+        np.argmax(actual, axis=1), np.argmax(np.asarray(expected), axis=1)
+    )
+
+    output.parent.mkdir(parents=True, exist_ok=True)
+    onnx.save(model, output)
+    model_bytes = output.read_bytes()
+    manifest = {
+        "format": "coolrl-lost-cities-match-onnx-v1",
+        "codename": codename,
+        "model_file": output.name,
+        "model_size_bytes": len(model_bytes),
+        "model_sha256": hashlib.sha256(model_bytes).hexdigest(),
+        "source_checkpoint": str(checkpoint),
+        "source_config": config.name,
+        "observation_size": MATCH_OBS_DIM,
+        "action_size": N_ACTIONS,
+        "hidden_size": cfg.network.hidden_size,
+        "num_layers": cfg.network.num_layers,
+        "dtype": "float32",
+        "validation_max_abs_error": float(np.max(np.abs(actual - np.asarray(expected)))),
+    }
+    output.with_suffix(".json").write_text(json.dumps(manifest, indent=2) + "\n")
+    print(f"exported {codename} -> {output} ({output.stat().st_size:,} bytes)")
+    print(
+        f"  sha256 {manifest['model_sha256'][:12]}  obs {MATCH_OBS_DIM}  max err "
+        f"{manifest['validation_max_abs_error']:.2e}"
+    )
+
+
+def main() -> None:
+    args = build_argparser().parse_args()
+    export_model(args.checkpoint, args.config, args.output, args.codename)
+
+
+if __name__ == "__main__":
+    main()

scripts/generate_match_parity_fixture.py

@@ -0,0 +1,97 @@
+#!/usr/bin/env python3
+"""Generate match states and their observations so TypeScript can be checked against JAX.
+
+The two observation builders must agree to the bit. A mismatch does not throw --
+the ONNX policy consumes a wrong vector quite happily and plays worse for reasons
+nobody can see. So the port is not trusted; it is checked.
+
+States are drawn from real random play so the fixture covers the awkward parts:
+mid-round, both seats to move, past a round roll-over, with a non-zero carry.
+"""
+
+from __future__ import annotations
+
+import json
+from pathlib import Path
+
+import jax
+import numpy as np
+
+from lost_cities_jax.match import MatchState, match_reset_from, match_step
+from lost_cities_jax.match_obs import MATCH_OBS_DIM, match_observation
+from lost_cities_jax.opponents import random_legal_action
+from lost_cities_jax.types import N_CARDS
+
+OUTPUT = Path(__file__).resolve().parents[1] / "web" / "src" / "game" / "match-parity-fixture.json"
+N_ROUNDS = 3
+
+
+def match_json(match: MatchState) -> dict:
+    round_state = match.round
+    return {
+        "round": {
+            "deckOrder": np.asarray(round_state.deck_order).astype(int).tolist(),
+            "drawPtr": int(round_state.draw_ptr),
+            "cardLoc": np.asarray(round_state.card_loc).astype(int).tolist(),
+            "handPublic": np.asarray(round_state.hand_public).astype(bool).tolist(),
+            "colTop": np.asarray(round_state.col_top).astype(int).tolist(),
+            "colHandshakes": np.asarray(round_state.col_hs).astype(int).tolist(),
+            "colLength": np.asarray(round_state.col_len).astype(int).tolist(),
+            "piles": [
+                np.asarray(round_state.pile[color, : int(round_state.pile_len[color])])
+                .astype(int)
+                .tolist()
+                for color in range(5)
+            ],
+            "toMove": int(round_state.to_move),
+            "stepCount": int(round_state.step_count),
+            "done": bool(round_state.done),
+        },
+        "deckOrders": np.asarray(match.deck_orders).astype(int).tolist(),
+        "coinFlips": np.asarray(match.coin_flips).astype(int).tolist(),
+        "roundIdx": int(match.round_idx),
+        "carry": np.asarray(match.carry).astype(int).tolist(),
+        "done": bool(match.done),
+    }
+
+
+def main() -> None:
+    rng = np.random.default_rng(20260715)
+    key = jax.random.PRNGKey(7)
+    rows = []
+
+    for match_index in range(6):
+        decks = np.stack([rng.permutation(N_CARDS) for _ in range(N_ROUNDS)])
+        coins = rng.integers(0, 2, size=(N_ROUNDS,))
+        match = match_reset_from(decks.astype(np.int8), coins.astype(np.int8))
+
+        # Sample the opening position and then every 17th ply, which lands in all
+        # three rounds and on both seats without hand-picking anything.
+        ply = 0
+        while not bool(match.done) and ply < 400:
+            if ply % 17 == 0 or ply == 0:
+                for player in (0, 1):
+                    obs = np.asarray(match_observation(match, player), dtype=np.float64)
+                    assert obs.shape == (MATCH_OBS_DIM,)
+                    rows.append(
+                        {
+                            "match": match_json(match),
+                            "player": player,
+                            "observation": [round(float(v), 7) for v in obs],
+                        }
+                    )
+            key, step_key = jax.random.split(key)
+            action = int(random_legal_action(match.round, match.round.to_move, step_key))
+            match, _, _ = match_step(match, action)
+            ply += 1
+        del match_index
+
+    OUTPUT.write_text(
+        json.dumps({"format": "jax-web-match-parity-v1", "obsDim": MATCH_OBS_DIM, "rows": rows})
+        + "\n"
+    )
+    print(f"wrote {len(rows)} rows -> {OUTPUT}")
+
+
+if __name__ == "__main__":
+    main()

scripts/serve_web_with_logs.py

@@ -0,0 +1,86 @@
+#!/usr/bin/env python3
+"""Serve the production web client and append completed games to JSONL."""
+
+from __future__ import annotations
+
+import argparse
+import json
+from http import HTTPStatus
+from http.server import SimpleHTTPRequestHandler, ThreadingHTTPServer
+from pathlib import Path
+from typing import Any
+
+MAX_RECORD_BYTES = 6_000_000
+
+# v2 adds the opponent's identity (codename + hash) and the match layer -- three
+# deals, the coin flips, and which round each move belongs to. v1 is still
+# accepted because 111 games were recorded under it; they are all altair, which
+# data/models.json records since v1 has nowhere to say so.
+SUPPORTED_FORMATS = frozenset({"lost-cities-web-game-v1", "lost-cities-web-game-v2"})
+
+
+def parse_record(body: bytes) -> dict[str, Any]:
+    if len(body) > MAX_RECORD_BYTES:
+        raise ValueError("record is too large")
+    value = json.loads(body)
+    if not isinstance(value, dict) or value.get("format") not in SUPPORTED_FORMATS:
+        raise ValueError("unsupported game record")
+    if not isinstance(value.get("gameId"), str) or not isinstance(value.get("moves"), list):
+        raise ValueError("invalid game record")
+    return value
+
+
+def make_handler(dist: Path, output: Path):
+    seen_ids: set[str] = set()
+    if output.exists():
+        for line in output.read_text(encoding="utf-8").splitlines():
+            try:
+                game_id = json.loads(line).get("gameId")
+                if isinstance(game_id, str):
+                    seen_ids.add(game_id)
+            except (json.JSONDecodeError, AttributeError):
+                continue
+
+    class Handler(SimpleHTTPRequestHandler):
+        def __init__(self, *args: Any, **kwargs: Any) -> None:
+            super().__init__(*args, directory=str(dist), **kwargs)
+
+        def do_POST(self) -> None:  # noqa: N802
+            if self.path != "/api/game-records":
+                self.send_error(HTTPStatus.NOT_FOUND)
+                return
+            try:
+                length = int(self.headers.get("content-length", "0"))
+                record = parse_record(self.rfile.read(length))
+            except (ValueError, json.JSONDecodeError) as error:
+                self.send_error(HTTPStatus.BAD_REQUEST, str(error))
+                return
+            game_id = record["gameId"]
+            if game_id not in seen_ids:
+                output.parent.mkdir(parents=True, exist_ok=True)
+                with output.open("a", encoding="utf-8") as stream:
+                    stream.write(
+                        json.dumps(record, ensure_ascii=False, separators=(",", ":")) + "\n"
+                    )
+                seen_ids.add(game_id)
+            self.send_response(HTTPStatus.NO_CONTENT)
+            self.end_headers()
+
+    return Handler
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--host", default="0.0.0.0")
+    parser.add_argument("--port", type=int, default=5173)
+    parser.add_argument("--dist", type=Path, default=Path("web/dist"))
+    parser.add_argument("--output", type=Path, default=Path("data/human-play/game-records.jsonl"))
+    args = parser.parse_args()
+    server = ThreadingHTTPServer((args.host, args.port), make_handler(args.dist, args.output))
+    print(f"Serving {args.dist} on http://{args.host}:{args.port}", flush=True)
+    print(f"Writing game records to {args.output}", flush=True)
+    server.serve_forever()
+
+
+if __name__ == "__main__":
+    main()

tests/lost_cities_jax/test_web_log_server.py

@@ -0,0 +1,34 @@
+import json
+
+import pytest
+
+from scripts.serve_web_with_logs import parse_record
+
+
+def test_parse_record_accepts_analysis_record():
+    record = {"format": "lost-cities-web-game-v1", "gameId": "game-1", "moves": []}
+    assert parse_record(json.dumps(record).encode()) == record
+
+
+@pytest.mark.parametrize("record", [{}, {"format": "other"}, {"format": "lost-cities-web-game-v1"}])
+def test_parse_record_rejects_invalid_input(record):
+    with pytest.raises(ValueError):
+        parse_record(json.dumps(record).encode())
+
+
+def test_parse_record_accepts_the_match_schema():
+    # v2 carries which model played and the three-round match; v1 had neither.
+    record = {
+        "format": "lost-cities-web-game-v2",
+        "gameId": "game-2",
+        "opponent": {"codename": "borealis", "hash": "13a25243de1c"},
+        "mode": 3,
+        "moves": [],
+    }
+    assert parse_record(json.dumps(record).encode()) == record
+
+
+def test_parse_record_still_accepts_v1():
+    # 111 games were recorded under v1; refusing them now would orphan them.
+    record = {"format": "lost-cities-web-game-v1", "gameId": "game-1", "moves": []}
+    assert parse_record(json.dumps(record).encode()) == record

web/.claude/skills/verify/SKILL.md

@@ -0,0 +1,68 @@
+---
+name: verify
+description: Drive the Lost Cities web client in a real browser to observe a change working — model loading, a full match, the result card, records being written.
+---
+
+# Verifying the web client
+
+The client is where a silent bug hides best. `npm test` and `tsc` were both green
+while the app was announcing the wrong winner and dropping every game record on
+the floor. Neither throws. Run it.
+
+## Build and serve
+
+```bash
+cd web && npm run build
+cd .. && tmux new-session -d -s websrv \
+  "uv run python scripts/serve_web_with_logs.py --host 127.0.0.1 --port 5199 \
+     --dist web/dist --output /tmp/verify-records.jsonl"
+```
+
+Serve the **built dist**, not the vite dev server — the deploy builds from dist,
+and the model is a static asset whose path only resolves there. Point `--output`
+at a scratch file so verification runs never touch `data/human-play/`.
+
+## Drive it
+
+Playwright is deliberately **not** a dependency: the `playwright` package downloads
+~114MB of Chromium on install, and `npm ci` runs in the deploy workflow. Install it
+for the run and uninstall after.
+
+```bash
+cd web
+npm i -D playwright --no-fund --no-audit && npx playwright install chromium
+# ... drive ...
+npm uninstall playwright
+```
+
+## Selectors that actually work
+
+Found by dumping the DOM; guessing at them wasted two runs.
+
+| What | Selector |
+|---|---|
+| a hand card | `button.card:not([disabled])` |
+| play onto an expedition | `button.lane__zone--mine.is-target` |
+| discard | `button.lane__discard.is-target` |
+| draw from deck | `button.deck-stack:not([disabled])` |
+| which model loaded | `.score-plaque--rival small` |
+| match progress | `.round-strip` |
+| final scores | `.result-card` |
+
+A turn is three clicks: pick a card, choose where it goes, then draw. The place and
+draw targets only appear **after** the card is selected, and only the legal ones are
+enabled — so click the card first, then query.
+
+The rival answers on a 620ms timer plus inference; ~200ms of slack between plies is
+enough. A full three-round match runs ~140 plies, so budget a few minutes.
+
+## Worth driving
+
+- **A whole match, not one round.** The round roll-over is where carry banks, and it
+  is where the result card got it wrong.
+- **Check the record actually saved.** A schema bump on the client silently 400s
+  against `scripts/serve_web_with_logs.py` until its allowlist is updated too.
+- **A stale save in localStorage.** Bump the key on a schema change; a half-migrated
+  save is worse than a fresh deal.
+- **The same seed twice.** In match mode the seed fixes all three deals and the coin
+  flips, so a match is a pure function of it.

web/BUGS.md

@@ -0,0 +1,21 @@
+# Web bug checklist
+
+- [ ] Prevent cards from launching more than once during initial loading.
+  - Implementation and automated checks complete; user visual acceptance pending.
+- [ ] Stack Hint, Undo, and Redo vertically at the bottom-right, aligned with the hand row.
+  - Implementation and automated checks complete; user visual acceptance pending.
+- [ ] Restore the latest game timeline and review position from local storage.
+  - Implementation and automated checks complete; user reload acceptance pending.
+- [ ] Keep Hint, Undo, Redo, and Play From Here in fixed vertical positions.
+  - Play From Here remains visible but disabled when unavailable; user visual acceptance pending.
+- [ ] Automatically record completed human-versus-AI games for later analysis.
+  - JSONL includes the full deal, both hands before every move, actions, draws, scores, and final state.
+  - Local production output: `data/human-play/game-records.jsonl` (was `runs/tmp/`, which is gitignored and documented as disposable).
+- [ ] Support game-record IDs when LAN HTTP does not expose `crypto.randomUUID`.
+  - Uses `crypto.getRandomValues` with a compatibility fallback; user acceptance pending.
+
+Verification:
+
+- `cd web && npm test`
+- `cd web && npm run build`
+- Visual acceptance is performed by the user; do not run browser automation.

web/public/models/borealis.json

@@ -0,0 +1,15 @@
+{
+  "format": "coolrl-lost-cities-match-onnx-v1",
+  "codename": "borealis",
+  "model_file": "borealis.onnx",
+  "model_size_bytes": 3327043,
+  "model_sha256": "13a25243de1c0c4f27dcac348b072cc6ea8ffd283dc033b877a37099fd809415",
+  "source_checkpoint": "runs/jax-ppo-match/2026-07-15_031529_match-scaled/latest",
+  "source_config": "match-selfplay.yaml",
+  "observation_size": 501,
+  "action_size": 96,
+  "hidden_size": 512,
+  "num_layers": 3,
+  "dtype": "float32",
+  "validation_max_abs_error": 9.1552734375e-05
+}

web/src/App.tsx

@@ -1,29 +1,53 @@
 import { useEffect, useMemo, useRef, useState } from "react";
 
-import { Board, type BoardHint } from "./components/Board";
+import { Board } from "./components/Board";
 import { Card, CardBack } from "./components/Card";
-import { ResultCard } from "./components/ResultCard";
-import { describeAction } from "./game/actions";
+import { ResultCard, RoundOverlay } from "./components/Scoreboard";
 import { cardColor, cardName } from "./game/cards";
 import {
   boardScore,
   currentHandSorted,
   decodeAction,
   encodeAction,
-  legalActionMask,
   previewPlacement,
-  resetFromOrder,
-  step,
 } from "./game/engine";
-import { deckOrderFromSeed, normalizeSeed, randomSeed } from "./game/random";
-import { DRAW_DECK, N_CARDS, PLAY, type GameState, type PlaceType } from "./game/types";
+import {
+  N_ROUNDS,
+  matchAdvanceRound,
+  matchFromOrders,
+  matchLegalActionMask,
+  matchScore,
+  matchStep,
+  roundScore,
+  type MatchState,
+} from "./game/match";
+import { MODEL_CODENAME, MODEL_HASH } from "./model/policy";
+import { matchFromSeed, normalizeSeed, randomSeed } from "./game/random";
+import { parseSavedGame, SAVED_GAME_KEY, type SavedGame } from "./game/persistence";
+import { DRAW_DECK, N_CARDS, PLAY, type PlaceType } from "./game/types";
 import { fallbackHeuristicPolicy, loadPolicy, type Policy } from "./model/policy";
 import { useCardMotion } from "./ui/useCardMotion";
 
-interface Hint {
-  action: number;
-  text: string;
-  probability: number | null;
+/** One deal, or the classic three-round match decided on the summed total. */
+export type Mode = 1 | 3;
+
+const MODE_KEY = "lost-cities.mode";
+
+function loadMode(): Mode {
+  const requested = new URLSearchParams(window.location.search).get("rounds");
+  if (requested === "3") return 3;
+  if (requested === "1") return 1;
+  // The classic three-round match is the default; a stored "1" opts down to a deal.
+  return window.localStorage.getItem(MODE_KEY) === "1" ? 1 : 3;
+}
+
+// Off by default: taking moves back is a training aid, not how the game is played,
+// and the default should be the honest game. When on, undo/redo walk the whole
+// timeline -- across the rival's moves and earlier rounds, not just your turn.
+const UNDO_KEY = "lost-cities.undo";
+
+function loadUndoEnabled(): boolean {
+  return window.localStorage.getItem(UNDO_KEY) === "on";
 }
 
 type Selection = { handSlot: number | null; placeType: PlaceType | null };
@@ -36,8 +60,9 @@ const EMPTY_SELECTION: Selection = { handSlot: null, placeType: null };
  * makes a finished game reviewable ply by ply.
  */
 interface Frame {
-  state: GameState;
+  state: MatchState;
   selection: Selection;
+  move?: { player: 0 | 1; action: number };
 }
 
 // The hand row is centered, so the binding constraint is the score plaque
@@ -66,15 +91,12 @@ function handCardStep(viewportWidth: number, count: number): number {
   return Math.max(minVisible, Math.min(relaxed, fitted));
 }
 
-/** A `?seed=` in the URL loads that exact deal, so a game can be shared or replayed. */
-function initialSeed(): string {
-  const fromUrl = new URLSearchParams(window.location.search).get("seed");
-  const seed = fromUrl === null ? "" : normalizeSeed(fromUrl);
-  return seed === "" ? randomSeed() : seed;
-}
-
-function openingFrame(seed: string): Frame {
-  return { state: resetFromOrder(deckOrderFromSeed(seed)), selection: EMPTY_SELECTION };
+/** A `?seed=` in the URL loads that exact deal, so a game can be shared or replayed.
+ *  In match mode the seed fixes all three deals and the coin flips too, so the
+ *  whole match is a pure function of it. */
+function openingFrame(seed: string, mode: Mode): Frame {
+  const { deckOrders, coinFlips } = matchFromSeed(seed);
+  return { state: matchFromOrders(deckOrders, coinFlips, mode), selection: EMPTY_SELECTION };
 }
 
 /** Keep the address bar in sync so the current deal stays shareable. */
@@ -84,29 +106,72 @@ function publishSeed(seed: string): void {
   window.history.replaceState(null, "", url);
 }
 
+function newGameId(): string {
+  if (typeof crypto.randomUUID === "function") return crypto.randomUUID();
+  if (typeof crypto.getRandomValues === "function") {
+    const bytes = crypto.getRandomValues(new Uint8Array(16));
+    bytes[6] = (bytes[6] & 0x0f) | 0x40;
+    bytes[8] = (bytes[8] & 0x3f) | 0x80;
+    const hex = Array.from(bytes, (byte) => byte.toString(16).padStart(2, "0")).join("");
+    return `${hex.slice(0, 8)}-${hex.slice(8, 12)}-${hex.slice(12, 16)}-${hex.slice(16, 20)}-${hex.slice(20)}`;
+  }
+  return `game-${Date.now()}-${Math.random().toString(36).slice(2)}`;
+}
+
+function initialGame(): Pick<SavedGame, "seed" | "frames" | "cursor" | "resultOpen" | "mode"> {
+  const requested = new URLSearchParams(window.location.search).get("seed");
+  const mode = loadMode();
+  const saved = parseSavedGame(window.localStorage.getItem(SAVED_GAME_KEY));
+  const requestedSeed = requested === null ? null : normalizeSeed(requested);
+  // A saved game only resumes into the mode it was played in.
+  if (saved && saved.mode === mode && (requestedSeed === null || requestedSeed === saved.seed)) {
+    return saved;
+  }
+  const seed = requestedSeed || randomSeed();
+  return { seed, mode, frames: [openingFrame(seed, mode)], cursor: 0, resultOpen: true };
+}
+
 function App() {
-  const [seed, setSeed] = useState<string>(initialSeed);
-  const [frames, setFrames] = useState<Frame[]>(() => [openingFrame(seed)]);
-  const [cursor, setCursor] = useState(0);
+  const [initial] = useState(initialGame);
+  const [seed, setSeed] = useState(initial.seed);
+  const [mode, setMode] = useState<Mode>(initial.mode as Mode);
+  const [frames, setFrames] = useState<Frame[]>(initial.frames);
+  const [cursor, setCursor] = useState(initial.cursor);
   const [seedDraft, setSeedDraft] = useState("");
-  const [resultOpen, setResultOpen] = useState(true);
+  const [resultOpen, setResultOpen] = useState(initial.resultOpen);
   const [policy, setPolicy] = useState<Policy | null>(null);
-  const [modelMessage, setModelMessage] = useState("LOADING FINAL PPO");
+  const [modelMessage, setModelMessage] = useState("BOREALIS 불러오는 중");
   const [thinking, setThinking] = useState(false);
   const [menuOpen, setMenuOpen] = useState(false);
   const [viewportWidth, setViewportWidth] = useState(() => window.innerWidth);
-  const [hint, setHint] = useState<Hint | null>(null);
-  const [hintPending, setHintPending] = useState(false);
+  const [undoEnabled, setUndoEnabled] = useState(loadUndoEnabled);
   const generation = useRef(0);
+  const gameId = useRef(newGameId());
+  const startedAt = useRef(new Date().toISOString());
+  const loggedGame = useRef<string | null>(null);
   const { cardRef, deckRef, overlayRef, resetMotion } = useCardMotion();
 
-  const { state, selection } = frames[cursor];
-  const canUndo = cursor > 0;
-  const canRedo = cursor < frames.length - 1;
-  const latestState = useRef(state);
+  const { state: match, selection } = frames[cursor];
+  // The board on screen is the round in play; the match is what decides the game.
+  const state = match.round;
+  // When enabled, undo/redo walk the whole timeline -- back through your committed
+  // moves, the rival's replies, and earlier rounds, not just the current turn. The
+  // rival is held whenever there are moves ahead to redo, so stepping back never
+  // makes it replay the move you are reviewing.
+  const canUndo = undoEnabled && cursor > 0;
+  const canRedo = undoEnabled && cursor < frames.length - 1;
+  const latestMatch = useRef(match);
 
   useEffect(() => { publishSeed(seed); }, [seed]);
-  useEffect(() => { latestState.current = state; }, [state]);
+  useEffect(() => { latestMatch.current = match; }, [match]);
+  useEffect(() => { window.localStorage.setItem(MODE_KEY, String(mode)); }, [mode]);
+  useEffect(() => {
+    window.localStorage.setItem(UNDO_KEY, undoEnabled ? "on" : "off");
+  }, [undoEnabled]);
+  useEffect(() => {
+    const saved: SavedGame = { version: 2, seed, mode, frames, cursor, resultOpen };
+    window.localStorage.setItem(SAVED_GAME_KEY, JSON.stringify(saved));
+  }, [cursor, frames, mode, resultOpen, seed]);
 
   useEffect(() => {
     function onResize() { setViewportWidth(window.innerWidth); }
@@ -116,7 +181,7 @@ function App() {
 
   const humanHand = useMemo(() => currentHandSorted(state, 0), [state]);
   const opponentHand = useMemo(() => currentHandSorted(state, 1), [state]);
-  const legal = useMemo(() => legalActionMask(state), [state]);
+  const legal = useMemo(() => matchLegalActionMask(match), [match]);
   const selectedCard = selection.handSlot === null ? null : humanHand[selection.handSlot];
   const selectedColor = selectedCard === null ? null : cardColor(selectedCard);
   const displayState = useMemo(
@@ -125,14 +190,23 @@ function App() {
       : previewPlacement(state, selection.handSlot, selection.placeType),
     [selection, state],
   );
-  const scores = useMemo(() => boardScore(displayState), [displayState]);
+  /** This round's board. */
+  const roundScores = useMemo(() => boardScore(displayState), [displayState]);
+  /** Rounds already banked plus the board in play — the number that decides a match. */
+  const scores = useMemo(
+    (): [number, number] => [
+      match.carry[0] + roundScores[0],
+      match.carry[1] + roundScores[1],
+    ],
+    [match.carry, roundScores],
+  );
 
   useEffect(() => {
     let cancelled = false;
     loadPolicy().then(({ policy: loaded, warning }) => {
       if (cancelled) return;
       setPolicy(loaded);
-      setModelMessage(warning ?? `${loaded.provider.toUpperCase()} · FINAL PPO`);
+      setModelMessage(warning ?? "BOREALIS");
     });
     return () => { cancelled = true; };
   }, []);
@@ -149,45 +223,64 @@ function App() {
   const rivalSuspended = canRedo;
 
   useEffect(() => {
-    if (!policy || rivalSuspended || state.done || state.toMove !== 1) return;
+    // No move while a finished round is being shown -- the player advances it.
+    if (!policy || rivalSuspended || match.done || match.roundComplete || state.toMove !== 1) return;
     const currentGeneration = generation.current;
     const timer = window.setTimeout(async () => {
       setThinking(true);
       try {
         let action: number;
         try {
-          action = await policy.action(state);
+          action = await policy.action(match);
         } catch (error) {
           console.error("AI action failed; falling back to heuristic policy", error);
           const fallback = fallbackHeuristicPolicy();
-          action = await fallback.action(state);
+          action = await fallback.action(match);
           if (generation.current !== currentGeneration) return;
           setPolicy(fallback);
-          setModelMessage("HEURISTIC FALLBACK (MODEL ERROR)");
+          setModelMessage("휴리스틱 대체 (모델 오류)");
         }
         if (generation.current !== currentGeneration) return;
-        pushFrame({ state: step(state, action), selection: EMPTY_SELECTION });
+        pushFrame({
+          state: matchStep(match, action),
+          selection: EMPTY_SELECTION,
+          move: { player: 1, action },
+        });
       } catch (error) {
         console.error("AI action failed even with heuristic fallback", error);
-        setModelMessage("MODEL INFERENCE ERROR");
+        setModelMessage("모델 추론 오류");
       } finally {
         if (generation.current === currentGeneration) setThinking(false);
       }
     }, 620);
     return () => window.clearTimeout(timer);
-  }, [policy, rivalSuspended, state]);
+  }, [policy, rivalSuspended, match, state.toMove]);
+
+  /** Deal the next round from the round-complete pause. */
+  function advanceRound() {
+    generation.current += 1;
+    resetMotion();
+    setFrames((items) => [...items.slice(0, cursor + 1), {
+      state: matchAdvanceRound(match),
+      selection: EMPTY_SELECTION,
+    }]);
+    setCursor(cursor + 1);
+  }
 
   /** Deal a game. Without a seed this rolls a fresh one; the same seed always
    *  reproduces the same deal, so `restart(seed)` also replays the current one. */
-  function restart(nextSeed: string = randomSeed()) {
+  function restart(nextSeed: string = randomSeed(), nextMode: Mode = mode) {
     generation.current += 1;
     resetMotion();
+    gameId.current = newGameId();
+    startedAt.current = new Date().toISOString();
+    loggedGame.current = null;
     setSeed(nextSeed);
-    setFrames([openingFrame(nextSeed)]);
+    setMode(nextMode);
+    setFrames([openingFrame(nextSeed, nextMode)]);
     setCursor(0);
     setSeedDraft("");
     setResultOpen(true);
-    setHint(null);
     setThinking(false);
     setMenuOpen(false);
   }
@@ -201,66 +294,29 @@ function App() {
     if (next < 0 || next >= frames.length) return;
     generation.current += 1;
     setCursor(next);
-    setHint(null);
     setThinking(false);
     setMenuOpen(false);
   }
 
-  const undo = () => moveCursor(cursor - 1);
-  const redo = () => moveCursor(cursor + 1);
-
-  /** Leave the reviewed position as the live one and let the rival play on. */
-  function resumeFromHere() {
-    setFrames((items) => items.slice(0, cursor + 1));
-    setMenuOpen(false);
-  }
-
-  /** Ask the policy driving the rival what it would do in your seat. */
-  async function requestHint() {
-    if (!policy || state.done || state.toMove !== 0 || hintPending) return;
-    const position = state;
-    setHintPending(true);
-    setMenuOpen(false);
-    try {
-      const ranked = await policy.rank(position).catch(async (error) => {
-        console.error("hint inference failed; falling back to heuristic", error);
-        return fallbackHeuristicPolicy().rank(position);
-      });
-      const [best] = ranked;
-      // The position can move on while inference runs — a hint for a stale
-      // position would point at the wrong hand slot.
-      if (best === undefined || latestState.current !== position) return;
-      setHint({
-        action: best.action,
-        text: describeAction(position, best.action),
-        probability: best.probability,
-      });
-    } catch (error) {
-      console.error("hint failed", error);
-    } finally {
-      setHintPending(false);
-    }
-  }
-
-  /** A hint answers one position; drop it as soon as anything is done or undone. */
-  useEffect(() => { setHint(null); }, [cursor, frames]);
+  const undo = () => { if (canUndo) moveCursor(cursor - 1); };
+  const redo = () => { if (canRedo) moveCursor(cursor + 1); };
 
   function chooseCard(handSlot: number) {
-    if (state.toMove !== 0 || state.done) return;
+    if (state.toMove !== 0 || match.done) return;
     const next = selection.handSlot !== handSlot || selection.placeType !== null
       ? { handSlot, placeType: null }
       : EMPTY_SELECTION;
-    pushFrame({ state, selection: next });
+    pushFrame({ state: match, selection: next });
   }
 
   function choosePlace(placeType: PlaceType) {
     if (selection.handSlot === null) return;
-    pushFrame({ state, selection: { ...selection, placeType } });
+    pushFrame({ state: match, selection: { ...selection, placeType } });
   }
 
   function cancelPlace() {
     if (selection.handSlot === null || selection.placeType === null) return;
-    pushFrame({ state, selection: { ...selection, placeType: null } });
+    pushFrame({ state: match, selection: { ...selection, placeType: null } });
   }
 
   useEffect(() => {
@@ -282,7 +338,11 @@ function App() {
   function commit(drawSource: number) {
     if (!legalDraw(drawSource) || selection.handSlot === null || selection.placeType === null) return;
     const action = encodeAction(selection.handSlot, selection.placeType, drawSource);
-    pushFrame({ state: step(state, action), selection: EMPTY_SELECTION });
+    pushFrame({
+      state: matchStep(match, action),
+      selection: EMPTY_SELECTION,
+      move: { player: 0, action },
+    });
   }
 
   // Cards actually drawn in the hand row (the selected card moves into the
@@ -299,44 +359,100 @@ function App() {
     (_, draw) => legal[encodeAction(selection.handSlot!, PLAY, draw)],
   ).some(Boolean);
 
-  const hintMove = hint === null ? null : decodeAction(hint.action);
-  const hintedCard = hintMove === null ? undefined : humanHand[hintMove.handSlot];
-  const boardHint: BoardHint | null = hintMove === null || hintedCard === undefined ? null : {
-    color: cardColor(hintedCard),
-    place: hintMove.placeType,
-    drawSource: hintMove.drawSource,
-  };
-  const hintConfidence = hint !== null && hint.probability !== null
-    ? `${Math.round(hint.probability * 100)}%`
-    : null;
-  const canHint = policy !== null && !state.done && state.toMove === 0;
 
   const outcome = scores[0] === scores[1]
-    ? "DRAW"
-    : scores[0] > scores[1] ? "YOU WIN" : "THE RIVAL WINS";
+    ? "무승부"
+    : scores[0] > scores[1] ? "승리" : "패배";
+
+  useEffect(() => {
+    if (!match.done || loggedGame.current === gameId.current) return;
+    loggedGame.current = gameId.current;
+    const moves = frames.flatMap((frame, index) => {
+      if (!frame.move || index === 0) return [];
+      const beforeMatch = frames[index - 1].state;
+      const afterMatch = frame.state;
+      const before = beforeMatch.round;
+      const after = afterMatch.round;
+      const decoded = decodeAction(frame.move.action);
+      const beforeHand = currentHandSorted(before, frame.move.player);
+      // A move that ends a round is followed by a fresh deal, so the drawn card
+      // has to be read from the hand it was drawn into, not from the next round's.
+      const rolledOver = afterMatch.roundIdx !== beforeMatch.roundIdx;
+      const afterHand = rolledOver ? beforeHand : currentHandSorted(after, frame.move.player);
+      return [{
+        ply: before.stepCount + 1,
+        round: beforeMatch.roundIdx + 1,
+        player: frame.move.player === 0 ? "human" : "ai",
+        action: frame.move.action,
+        playedCard: beforeHand[decoded.handSlot],
+        placeType: decoded.placeType === PLAY ? "play" : "discard",
+        drawSource: decoded.drawSource === DRAW_DECK ? "deck" : `discard_${decoded.drawSource - 1}`,
+        drawnCard: rolledOver
+          ? null
+          : afterHand.find((card) => !beforeHand.includes(card)) ?? null,
+        handsBefore: {
+          human: currentHandSorted(before, 0),
+          ai: currentHandSorted(before, 1),
+        },
+        scoresAfter: matchScore(afterMatch),
+      }];
+    });
+    const record = {
+      format: "lost-cities-web-game-v2",
+      gameId: gameId.current,
+      seed,
+      // Which model actually played. The old records stored the on-screen label,
+      // which stops identifying anything the moment there is a second model.
+      opponent: { codename: MODEL_CODENAME, hash: MODEL_HASH },
+      mode,
+      startedAt: startedAt.current,
+      finishedAt: new Date().toISOString(),
+      deckOrders: match.deckOrders.slice(0, mode),
+      coinFlips: match.coinFlips.slice(0, mode),
+      moves,
+      finalScores: scores,
+      roundScores: match.roundHistory,
+      outcome,
+      finalState: match,
+      policy: modelMessage,
+    };
+    void fetch("./api/game-records", {
+      method: "POST",
+      headers: { "content-type": "application/json" },
+      body: JSON.stringify(record),
+      keepalive: true,
+    }).then((response) => {
+      if (!response.ok) throw new Error(`game log HTTP ${response.status}`);
+    }).catch((error) => {
+      loggedGame.current = null;
+      console.warn("Game record was not saved", error);
+    });
+  }, [frames, match, mode, modelMessage, outcome, scores, seed, state.done]);
   // While the rival is paused on its own turn (or on the final position) nothing
   // will happen until the move is redone or play is resumed — say so. On your own
   // turn the normal prompt still applies: you can simply play on from here.
-  const reviewingRival = rivalSuspended && (state.done || state.toMove === 1);
+  const reviewingRival = rivalSuspended && (match.done || state.toMove === 1);
   const status = reviewingRival
-    ? `Reviewing ply ${state.stepCount} — redo, or play on from here`
-    : state.done
+    ? "되돌리는 중 — 다시 진행하려면 앞으로"
+    : match.done
       ? outcome
+      : match.roundComplete
+        ? `${match.roundIdx + 1}라운드 종료`
         : state.toMove === 1
-        ? thinking ? "The rival is thinking ···" : "The rival's turn"
+          ? thinking ? "상대가 생각 중이에요" : "상대 차례"
           : selection.handSlot === null
-          ? "Play or discard a card"
+            ? "낼 카드나 버릴 카드를 고르세요"
             : selection.placeType === null
-            ? `Choose a destination for ${cardName(selectedCard!)}`
-            : "Draw — deck or a discard pile";
+              ? `${cardName(selectedCard!)} 놓을 곳을 고르세요`
+              : "카드를 뽑으세요 — 덱 또는 버림패";
 
   return (
     <main className="app-shell">
-      <section className="score-plaque score-plaque--rival" aria-label={`Rival score ${scores[1]}`}>
-        <div><strong>THE RIVAL</strong><small>{modelMessage}</small></div><b>{scores[1]}</b>
+      <section className="score-plaque score-plaque--rival" aria-label={`상대 점수 ${scores[1]}`}>
+        <div><strong>상대</strong><small>{modelMessage}</small></div><b>{scores[1]}</b>
       </section>
 
-      <div className="opponent-hand" aria-label="Rival hand">
+      <div className="opponent-hand" aria-label="상대 손패">
         {opponentHand.map((card) => (
           state.handPublic[card]
             ? <Card mini card={card} key={card} innerRef={cardRef(card)} />
@@ -345,28 +461,64 @@ function App() {
       </div>
 
       <div className="menu-wrap">
-        <button className="menu-button" onClick={() => setMenuOpen((open) => !open)}>MENU</button>
+        <button className="menu-button" onClick={() => setMenuOpen((open) => !open)}>메뉴</button>
         {menuOpen && (
           <div className="menu-popover">
-            <button onClick={() => restart()}>NEW GAME</button>
-            <button onClick={() => restart(seed)}>REPLAY THIS SEED</button>
+            <button onClick={() => restart()}>새 게임</button>
+            <button onClick={() => restart(seed)}>같은 시드로 다시</button>
+            <div className="menu-modes" role="group" aria-label="게임 길이">
+              <button
+                className={mode === 3 ? "is-active" : ""}
+                onClick={() => restart(randomSeed(), 3)}
+              >
+                3라운드 매치
+              </button>
+              <button
+                className={mode === 1 ? "is-active" : ""}
+                onClick={() => restart(randomSeed(), 1)}
+              >
+                단판
+              </button>
+            </div>
             <label className="menu-seed">
-              <span>DEAL SEED · {seed}</span>
+              <span>시드 · {seed}</span>
               <input
                 value={seedDraft}
-                placeholder="load a deal seed…"
+                placeholder="시드 입력…"
                 onChange={(event) => setSeedDraft(event.target.value)}
                 onKeyDown={(event) => { if (event.key === "Enter") loadSeed(); }}
-                aria-label="Load a deal by seed"
+                aria-label="시드로 딜 불러오기"
               />
             </label>
-            <button onClick={loadSeed} disabled={normalizeSeed(seedDraft) === ""}>DEAL THIS SEED</button>
-            <span>THE SEED SHUFFLES THE DECK ONLY · THE POLICY IS DETERMINISTIC</span>
-            <span>{policy ? `${policy.provider.toUpperCase()} POLICY` : "LOADING POLICY"}</span>
+            <button onClick={loadSeed} disabled={normalizeSeed(seedDraft) === ""}>이 시드로 시작</button>
+            <label className="menu-toggle">
+              <input
+                type="checkbox"
+                checked={undoEnabled}
+                onChange={(event) => setUndoEnabled(event.target.checked)}
+              />
+              <span>되돌리기 / 다시 · 전체 되감기</span>
+            </label>
           </div>
         )}
       </div>
 
+      {mode === N_ROUNDS && (
+        <div className="round-strip" aria-label={`${N_ROUNDS}라운드 중 ${match.roundIdx + 1}라운드`}>
+          {Array.from({ length: N_ROUNDS }, (_, round) => (
+            <span
+              key={round}
+              className={
+                round < match.roundIdx ? "is-done" : round === match.roundIdx ? "is-live" : ""
+              }
+            >
+              {round + 1}
+            </span>
+          ))}
+          <em>누적 {match.carry[0]} : {match.carry[1]}</em>
+        </div>
+      )}
+
       <div className="table-center">
         <Board
           state={displayState}
@@ -374,7 +526,6 @@ function App() {
           selectedPlace={selection.placeType}
           canPlay={canPlaySelected}
           canDraw={legalDraw}
-          hint={boardHint}
           cardRef={cardRef}
           onChoosePlace={choosePlace}
           onCancelPlace={cancelPlace}
@@ -384,38 +535,29 @@ function App() {
 
       <button
         type="button"
-        className={`deck-stack ${legalDraw(DRAW_DECK) ? "is-draw-target" : ""} ${hintMove?.drawSource === DRAW_DECK ? "is-hinted" : ""}`}
+        className={`deck-stack ${legalDraw(DRAW_DECK) ? "is-draw-target" : ""}`}
         disabled={!legalDraw(DRAW_DECK)}
         onClick={() => commit(DRAW_DECK)}
-        aria-label={`Draw from deck, ${N_CARDS - state.drawPtr} cards left`}
+        aria-label={`덱에서 뽑기, ${N_CARDS - state.drawPtr}장 남음`}
       >
         <CardBack innerRef={deckRef} />
         <strong>{N_CARDS - state.drawPtr}</strong>
-        <span>CARDS LEFT</span>
+        <span>남은 카드</span>
       </button>
 
       <div className="prompt-row">
         <p className={`turn-prompt ${thinking ? "turn-prompt--thinking" : ""} ${reviewingRival ? "turn-prompt--review" : ""}`}>
-          {hint ? `Hint — ${hint.text}${hintConfidence ? ` (${hintConfidence})` : ""}` : status}
+          {status}
         </p>
-        <button
-          type="button"
-          className={`hint-button ${hint ? "is-active" : ""}`}
-          onClick={requestHint}
-          disabled={!canHint || hintPending}
-        >
-          {hintPending ? "THINKING…" : "HINT"}
-        </button>
       </div>
 
-      <section className="human-hand" aria-label="Your hand">
+      <section className="human-hand" aria-label="내 손패">
         {renderedHand.map(({ card, slot }, position) => (
           <Card
             card={card}
             key={card}
             selected={selection.handSlot === slot}
-            hinted={hintedCard === card}
-            disabled={state.toMove !== 0 || state.done}
+            disabled={state.toMove !== 0 || match.done}
             onClick={() => chooseCard(slot)}
             style={position === 0 ? undefined : { marginLeft: handCardMarginLeft }}
             innerRef={cardRef(card)}
@@ -423,37 +565,52 @@ function App() {
         ))}
       </section>
 
-      <section className="score-plaque score-plaque--human" aria-label={`Your score ${scores[0]}`}>
-        <div><strong>YOU <i /></strong><small>{state.toMove === 0 && !state.done ? "YOUR TURN" : "EXPEDITION LEAD"}</small></div><b>{scores[0]}</b>
+      <section className="score-plaque score-plaque--human" aria-label={`내 점수 ${scores[0]}`}>
+        <div><strong>나 <i /></strong><small>{state.toMove === 0 && !match.done ? "내 차례" : "현재 점수"}</small></div><b>{scores[0]}</b>
       </section>
 
-      <div className="history-bar">
-        <button type="button" onClick={undo} disabled={!canUndo} aria-label="Undo one action" title="Undo (←)">
-          ↶ <span>UNDO</span>
+      <div className="control-stack">
+        {undoEnabled && (
+          <>
+            <button type="button" onClick={undo} disabled={!canUndo} aria-label="한 단계 되돌리기" title="되돌리기 (←)">
+              ↶ <span>되돌리기</span>
             </button>
-        <button type="button" onClick={redo} disabled={!canRedo} aria-label="Redo one action" title="Redo (→)">
-          ↷ <span>REDO</span>
-        </button>
-        {rivalSuspended && (
-          <button type="button" className="history-bar__resume" onClick={resumeFromHere}>
-            PLAY FROM HERE
+            <button type="button" onClick={redo} disabled={!canRedo} aria-label="한 단계 다시" title="다시 (→)">
+              ↷ <span>다시</span>
             </button>
+          </>
         )}
-        {state.done && !resultOpen && (
-          <button type="button" onClick={() => setResultOpen(true)}>SCORE</button>
-        )}
+        <button
+          type="button"
+          className={`control-stack__score ${match.done && !resultOpen ? "is-available" : ""}`}
+          onClick={() => setResultOpen(true)}
+          disabled={!match.done || resultOpen}
+          aria-hidden={!match.done || resultOpen}
+        >
+          결과
+        </button>
       </div>
 
       {/* Cards in flight back to the deck (undo of a draw) are re-parented here
           after React has already removed them from the table. */}
       <div className="motion-overlay" ref={overlayRef} aria-hidden="true" />
 
-      {state.done && resultOpen && (
+      {match.roundComplete && (
+        <RoundOverlay
+          state={state}
+          roundIdx={match.roundIdx}
+          totalRounds={mode}
+          matchTotals={scores}
+          onContinue={advanceRound}
+        />
+      )}
+
+      {match.done && resultOpen && (
         <ResultCard
           state={state}
           outcome={outcome}
-          seed={seed}
-          onReview={() => setResultOpen(false)}
+          roundHistory={mode > 1 ? match.roundHistory : [roundScore(match)]}
+          totalRounds={mode}
           onPlayAgain={() => restart()}
         />
       )}

web/src/components/Board.tsx

@@ -9,20 +9,12 @@ import {
 import { cardsOnBoard } from "../game/engine";
 import { DISCARD, PLAY, type GameState, type PlaceType } from "../game/types";
 
-/** Where the hint says to put the card and where to draw from, if hinting. */
-export interface BoardHint {
-  color: number;
-  place: PlaceType;
-  drawSource: number;
-}
-
 interface BoardProps {
   state: GameState;
   selectedColor: number | null;
   selectedPlace: PlaceType | null;
   canPlay: boolean;
   canDraw: (source: number) => boolean;
-  hint: BoardHint | null;
   cardRef: (card: number) => (element: HTMLElement | null) => void;
   onChoosePlace: (place: PlaceType) => void;
   onCancelPlace: () => void;
@@ -58,7 +50,6 @@ export function Board({
   selectedPlace,
   canPlay,
   canDraw,
-  hint,
   cardRef,
   onChoosePlace,
   onCancelPlace,
@@ -75,9 +66,6 @@ export function Board({
         const drawTarget = selectedPlace !== null && canDraw(color + 1);
         const discardChosen = selectedPlace === DISCARD && selectedColor === color;
         const playChosen = selectedPlace === PLAY && selectedColor === color;
-        const hintPlay = hint?.place === PLAY && hint.color === color;
-        const hintDiscard = hint?.place === DISCARD && hint.color === color;
-        const hintDraw = hint?.drawSource === color + 1;
         return (
           <article
             className="lane"
@@ -97,7 +85,7 @@ export function Board({
 
             <button
               type="button"
-              className={`lane__discard ${discardTarget ? "is-target" : ""} ${drawTarget ? "is-draw-target" : ""} ${discardChosen ? "is-chosen" : ""} ${hintDiscard || hintDraw ? "is-hinted" : ""}`}
+              className={`lane__discard ${discardTarget ? "is-target" : ""} ${drawTarget ? "is-draw-target" : ""} ${discardChosen ? "is-chosen" : ""}`}
               onClick={() => drawTarget ? onDraw(color + 1) : discardChosen ? onCancelPlace() : discardTarget ? onChoosePlace(DISCARD) : undefined}
               disabled={!discardTarget && !drawTarget && !discardChosen}
               aria-label={discardChosen ? `Cancel discarding to ${name} pile` : `${name} discard pile`}
@@ -113,7 +101,7 @@ export function Board({
 
             <button
               type="button"
-              className={`lane__zone lane__zone--mine ${playTarget ? "is-target" : ""} ${playChosen ? "is-chosen" : ""} ${hintPlay ? "is-hinted" : ""}`}
+              className={`lane__zone lane__zone--mine ${playTarget ? "is-target" : ""} ${playChosen ? "is-chosen" : ""}`}
               onClick={() => playChosen ? onCancelPlace() : playTarget ? onChoosePlace(PLAY) : undefined}
               disabled={!playTarget && !playChosen}
               aria-label={playChosen ? `Cancel playing to your ${name} expedition` : `Your ${name} expedition`}

web/src/components/Card.tsx

@@ -14,7 +14,6 @@ interface CardProps {
    *  hand row is only as tall as a card back, so a full card would burst it. */
   mini?: boolean;
   selected?: boolean;
-  hinted?: boolean;
   disabled?: boolean;
   onClick?: () => void;
   style?: React.CSSProperties;
@@ -26,7 +25,6 @@ export function Card({
   compact = false,
   mini = false,
   selected = false,
-  hinted = false,
   disabled = false,
   onClick,
   style: styleOverride,
@@ -35,7 +33,7 @@ export function Card({
   const color = cardColor(card);
   const value = cardValue(card);
   const handshake = isHandshake(card);
-  const className = `card ${compact ? "card--compact" : ""} ${mini ? "card--mini" : ""} ${selected ? "card--selected" : ""} ${hinted ? "card--hinted" : ""}`;
+  const className = `card ${compact ? "card--compact" : ""} ${mini ? "card--mini" : ""} ${selected ? "card--selected" : ""}`;
   const style = { "--card-color": COLOR_HEX[color], ...styleOverride } as React.CSSProperties;
   const content = (
     <>

web/src/components/ResultCard.tsx

@@ -1,91 +0,0 @@
-import { COLOR_GLYPHS, COLOR_HEX, EXPEDITION_NAMES } from "../game/cards";
-import { scoreBreakdown } from "../game/scoring";
-import type { GameState } from "../game/types";
-
-interface ResultCardProps {
-  state: GameState;
-  outcome: string;
-  seed: string;
-  onReview: () => void;
-  onPlayAgain: () => void;
-}
-
-function signed(value: number): string {
-  return value > 0 ? `+${value}` : String(value);
-}
-
-export function ResultCard({ state, outcome, seed, onReview, onPlayAgain }: ResultCardProps) {
-  const you = scoreBreakdown(state, 0);
-  const rival = scoreBreakdown(state, 1);
-
-  return (
-    <div className="result-overlay">
-      <section className="result-card" aria-label="Final score">
-        <header className="result-card__head">
-          <p>ROUND COMPLETE</p>
-          <h1>{outcome}</h1>
-          <strong>{you.total} <i>:</i> {rival.total}</strong>
-        </header>
-
-        <table className="result-table">
-          <thead>
-            <tr>
-              <th scope="col">EXPEDITION</th>
-              <th scope="col" colSpan={2}>YOU</th>
-              <th scope="col" colSpan={2}>THE RIVAL</th>
-            </tr>
-            <tr className="result-table__subhead">
-              <th scope="col" />
-              <th scope="col">CARDS</th>
-              <th scope="col">SCORE</th>
-              <th scope="col">CARDS</th>
-              <th scope="col">SCORE</th>
-            </tr>
-          </thead>
-          <tbody>
-            {EXPEDITION_NAMES.map((name, color) => {
-              const mine = you.colors[color];
-              const theirs = rival.colors[color];
-              return (
-                <tr key={name}>
-                  <th scope="row" style={{ "--lane-color": COLOR_HEX[color] } as React.CSSProperties}>
-                    <i>{COLOR_GLYPHS[color]}</i>{name}
-                  </th>
-                  <td className="result-table__detail">
-                    {mine.cards ? `${mine.cards}${mine.handshakes ? ` ·×${mine.handshakes + 1}` : ""}` : "—"}
-                  </td>
-                  <td className={mine.score < 0 ? "is-negative" : mine.score > 0 ? "is-positive" : ""}>
-                    {mine.cards ? signed(mine.score) : "—"}
-                  </td>
-                  <td className="result-table__detail">
-                    {theirs.cards ? `${theirs.cards}${theirs.handshakes ? ` ·×${theirs.handshakes + 1}` : ""}` : "—"}
-                  </td>
-                  <td className={theirs.score < 0 ? "is-negative" : theirs.score > 0 ? "is-positive" : ""}>
-                    {theirs.cards ? signed(theirs.score) : "—"}
-                  </td>
-                </tr>
-              );
-            })}
-          </tbody>
-          <tfoot>
-            <tr>
-              <th scope="row">TOTAL</th>
-              <td className="result-table__detail" />
-              <td>{signed(you.total)}</td>
-              <td className="result-table__detail" />
-              <td>{signed(rival.total)}</td>
-            </tr>
-          </tfoot>
-        </table>
-
-        <footer className="result-card__foot">
-          <small>DEAL SEED {seed}</small>
-          <div>
-            <button type="button" onClick={onReview}>REVIEW GAME</button>
-            <button type="button" className="is-primary" onClick={onPlayAgain}>PLAY AGAIN</button>
-          </div>
-        </footer>
-      </section>
-    </div>
-  );
-}

web/src/components/Scoreboard.tsx

@@ -0,0 +1,167 @@
+import { COLOR_GLYPHS, COLOR_HEX, EXPEDITION_NAMES_KO } from "../game/cards";
+import { scoreBreakdown } from "../game/scoring";
+import type { GameState } from "../game/types";
+
+function signed(value: number): string {
+  return value > 0 ? `+${value}` : String(value);
+}
+
+function scoreClass(value: number): string {
+  return value < 0 ? "is-negative" : value > 0 ? "is-positive" : "";
+}
+
+/** Per-expedition breakdown of one round's board: cards (with the wager multiplier)
+ *  and score, for both players. */
+function ColorTable({ state }: { state: GameState }) {
+  const you = scoreBreakdown(state, 0);
+  const rival = scoreBreakdown(state, 1);
+  return (
+    <table className="result-table">
+      <thead>
+        <tr>
+          <th scope="col">원정</th>
+          <th scope="col" colSpan={2}>나</th>
+          <th scope="col" colSpan={2}>상대</th>
+        </tr>
+        <tr className="result-table__subhead">
+          <th scope="col" />
+          <th scope="col">카드</th>
+          <th scope="col">점수</th>
+          <th scope="col">카드</th>
+          <th scope="col">점수</th>
+        </tr>
+      </thead>
+      <tbody>
+        {EXPEDITION_NAMES_KO.map((name, color) => {
+          const mine = you.colors[color];
+          const theirs = rival.colors[color];
+          const cards = (c: { cards: number; handshakes: number }) =>
+            c.cards ? `${c.cards}${c.handshakes ? ` ·×${c.handshakes + 1}` : ""}` : "—";
+          return (
+            <tr key={name}>
+              <th scope="row" style={{ "--lane-color": COLOR_HEX[color] } as React.CSSProperties}>
+                <i>{COLOR_GLYPHS[color]}</i>{name}
+              </th>
+              <td className="result-table__detail">{cards(mine)}</td>
+              <td className={scoreClass(mine.score)}>{mine.cards ? signed(mine.score) : "—"}</td>
+              <td className="result-table__detail">{cards(theirs)}</td>
+              <td className={scoreClass(theirs.score)}>{theirs.cards ? signed(theirs.score) : "—"}</td>
+            </tr>
+          );
+        })}
+      </tbody>
+      <tfoot>
+        <tr>
+          <th scope="row">합계</th>
+          <td className="result-table__detail" />
+          <td>{signed(you.total)}</td>
+          <td className="result-table__detail" />
+          <td>{signed(rival.total)}</td>
+        </tr>
+      </tfoot>
+    </table>
+  );
+}
+
+/** The pause between rounds: the finished round's per-colour scores and where the
+ *  match stands, held until the player chooses to deal the next round. */
+export function RoundOverlay({
+  state,
+  roundIdx,
+  totalRounds,
+  matchTotals,
+  onContinue,
+}: {
+  state: GameState;
+  roundIdx: number;
+  totalRounds: number;
+  matchTotals: [number, number];
+  onContinue: () => void;
+}) {
+  return (
+    <div className="result-overlay">
+      <section className="result-card" aria-label={`${roundIdx + 1}라운드 결과`}>
+        <header className="result-card__head">
+          <p>{roundIdx + 1}라운드 종료</p>
+          <strong>{matchTotals[0]} <i>:</i> {matchTotals[1]}</strong>
+          <small className="result-card__banked">누적 점수 · 나 {matchTotals[0]} · 상대 {matchTotals[1]}</small>
+        </header>
+        <ColorTable state={state} />
+        <footer className="result-card__foot">
+          <button type="button" className="is-primary" onClick={onContinue}>
+            {roundIdx + 1 >= totalRounds - 1 ? "마지막 라운드로" : `${roundIdx + 2}라운드 시작`}
+          </button>
+        </footer>
+      </section>
+    </div>
+  );
+}
+
+/** The end of the game: the outcome, the final round's per-colour breakdown, and --
+ *  for a match -- a single table summarising all three rounds. */
+export function ResultCard({
+  state,
+  outcome,
+  roundHistory,
+  totalRounds,
+  onPlayAgain,
+}: {
+  state: GameState;
+  outcome: string;
+  /** Each finished round's [you, rival] total, including this last one. */
+  roundHistory: [number, number][];
+  totalRounds: number;
+  onPlayAgain: () => void;
+}) {
+  const isMatch = totalRounds > 1;
+  const totals = roundHistory.reduce(
+    (acc, [you, rival]) => [acc[0] + you, acc[1] + rival] as [number, number],
+    [0, 0] as [number, number],
+  );
+
+  return (
+    <div className="result-overlay">
+      <section className="result-card" aria-label="최종 결과">
+        <header className="result-card__head">
+          <p>{isMatch ? `${totalRounds}라운드 매치 종료` : "게임 종료"}</p>
+          <h1>{outcome}</h1>
+          <strong>{totals[0]} <i>:</i> {totals[1]}</strong>
+        </header>
+
+        {isMatch ? (
+          <table className="result-table result-table--rounds">
+            <thead>
+              <tr>
+                <th scope="col">라운드</th>
+                <th scope="col">나</th>
+                <th scope="col">상대</th>
+              </tr>
+            </thead>
+            <tbody>
+              {roundHistory.map(([you, rival], round) => (
+                <tr key={round}>
+                  <th scope="row">{round + 1}라운드</th>
+                  <td className={scoreClass(you)}>{signed(you)}</td>
+                  <td className={scoreClass(rival)}>{signed(rival)}</td>
+                </tr>
+              ))}
+            </tbody>
+            <tfoot>
+              <tr className="result-table__match">
+                <th scope="row">합계</th>
+                <td>{signed(totals[0])}</td>
+                <td>{signed(totals[1])}</td>
+              </tr>
+            </tfoot>
+          </table>
+        ) : (
+          <ColorTable state={state} />
+        )}
+
+        <footer className="result-card__foot">
+          <button type="button" className="is-primary" onClick={onPlayAgain}>다시 하기</button>
+        </footer>
+      </section>
+    </div>
+  );
+}

web/src/game/cards.ts

@@ -2,6 +2,7 @@ import { CARDS_PER_COLOR } from "./types";
 
 export const COLOR_NAMES = ["Red", "Blue", "Green", "Gold", "Violet"] as const;
 export const EXPEDITION_NAMES = ["Volcano", "Ocean", "Jungle", "Desert", "Cavern"] as const;
+export const EXPEDITION_NAMES_KO = ["화산", "바다", "정글", "사막", "동굴"] as const;
 export const COLOR_HEX = ["#b94737", "#3569a7", "#397b4e", "#b37d20", "#704487"] as const;
 export const COLOR_GLYPHS = ["▲", "≋", "♧", "◆", "⬟"] as const;
 

web/src/game/engine.ts

@@ -32,7 +32,11 @@ export function shuffledDeck(random: () => number = Math.random): number[] {
   return deck;
 }
 
-export function resetFromOrder(deckOrder: number[]): GameState {
+/**
+ * `firstPlayer` moves first. Rounds two and three of a classic match are led by
+ * whoever is ahead on points, so the match layer sets this per round.
+ */
+export function resetFromOrder(deckOrder: number[], firstPlayer: 0 | 1 | number = 0): GameState {
   if (deckOrder.length !== N_CARDS || new Set(deckOrder).size !== N_CARDS) {
     throw new Error("deckOrder must be a permutation of 0..59");
   }
@@ -48,7 +52,7 @@ export function resetFromOrder(deckOrder: number[]): GameState {
     colHandshakes: emptyMatrix(2, N_COLORS),
     colLength: emptyMatrix(2, N_COLORS),
     piles: Array.from({ length: N_COLORS }, () => []),
-    toMove: 0,
+    toMove: (firstPlayer === 1 ? 1 : 0) as 0 | 1,
     stepCount: 0,
     done: false,
   };

web/src/game/match.ts

@@ -0,0 +1,146 @@
+import { boardScore, legalActionMask, resetFromOrder, step } from "./engine";
+import type { GameState } from "./types";
+
+export const N_ROUNDS = 3;
+
+/**
+ * Classic Lost Cities: three rounds, scores summed, highest total wins.
+ *
+ * The single-round engine is left alone -- it is the rules oracle the Python
+ * engine is differential-tested against. Only two rules live up here, both from
+ * the Kosmos rulebook:
+ *
+ *   "If after three games you have the highest overall score, you win."
+ *   "The player who has more points begins" the next game -- not alternating.
+ *
+ * The rulebook says nothing about an exact tie, so the starter falls back to a
+ * coin flip, drawn up front with the deals. This mirrors src/lost_cities_jax/match.py;
+ * the two are checked against each other in match.test.ts.
+ */
+export interface MatchState {
+  round: GameState;
+  /** Every deal of the match, shuffled up front. */
+  deckOrders: number[][];
+  /** Tie-breaking starters, used only when the scores are level. */
+  coinFlips: number[];
+  roundIdx: number;
+  /** Points banked by each player in the rounds already finished. */
+  carry: [number, number];
+  /**
+   * 1 for a one-off deal, 3 for the classic match.
+   *
+   * The observation always reports the round index out of three regardless --
+   * that is the space the policy was trained on, and a one-off deal is simply its
+   * round one, played at a carry of zero.
+   */
+  totalRounds: number;
+  /**
+   * The round's deck is exhausted and its board is being shown, but the match has
+   * not rolled into the next round yet. The engine advances rounds atomically; the
+   * client holds here so the finished board and its per-colour scores are visible
+   * before `matchAdvanceRound` deals the next one. Never set on the last round --
+   * that sets `done` instead.
+   */
+  roundComplete: boolean;
+  /** Each finished round's [you, rival] total, in order, for the final summary. */
+  roundHistory: [number, number][];
+  done: boolean;
+}
+
+/**
+ * Whoever has banked more points leads; level scores fall back to the coin.
+ *
+ * Round one needs no special case: carry is (0, 0) there, so the tie branch
+ * already picks the coin flip, which is exactly the rulebook's arbitrary
+ * "oldest player begins".
+ */
+export function startingPlayer(
+  carry: readonly [number, number],
+  roundIdx: number,
+  coinFlips: readonly number[],
+): number {
+  const lead = carry[0] - carry[1];
+  if (lead > 0) return 0;
+  if (lead < 0) return 1;
+  return coinFlips[roundIdx];
+}
+
+export function matchFromOrders(
+  deckOrders: number[][],
+  coinFlips: number[],
+  totalRounds: number = N_ROUNDS,
+): MatchState {
+  const carry: [number, number] = [0, 0];
+  return {
+    round: resetFromOrder(deckOrders[0], startingPlayer(carry, 0, coinFlips)),
+    deckOrders,
+    coinFlips,
+    roundIdx: 0,
+    carry,
+    totalRounds,
+    roundComplete: false,
+    roundHistory: [],
+    done: false,
+  };
+}
+
+/** Running totals: rounds already banked plus the board in play. */
+export function matchScore(state: MatchState): [number, number] {
+  const board = boardScore(state.round);
+  return [state.carry[0] + board[0], state.carry[1] + board[1]];
+}
+
+export function matchLegalActionMask(state: MatchState): boolean[] {
+  const mask = legalActionMask(state.round);
+  // No moves while the match is over or a finished round is being shown.
+  return state.done || state.roundComplete ? mask.map(() => false) : mask;
+}
+
+/** Score of the round currently on the board, by player. */
+export function roundScore(state: MatchState): [number, number] {
+  return boardScore(state.round);
+}
+
+/**
+ * Play one ply. When the deck runs out this holds on the finished board
+ * (`roundComplete`) rather than rolling straight into the next round, so the
+ * client can show the round's scores; `matchAdvanceRound` continues. The last
+ * round sets `done` instead. Rounds one and two only bank into `carry` -- the
+ * summed total decides the match.
+ */
+export function matchStep(state: MatchState, action: number): MatchState {
+  const played = step(state.round, action);
+  if (!played.done || state.done || state.roundComplete) {
+    return { ...state, round: played };
+  }
+
+  const board = boardScore(played);
+  if (state.roundIdx >= state.totalRounds - 1) {
+    return {
+      ...state,
+      round: played,
+      roundHistory: [...state.roundHistory, board],
+      done: true,
+    };
+  }
+  return { ...state, round: played, roundComplete: true };
+}
+
+/** Bank the finished round and deal the next one. No-op unless roundComplete. */
+export function matchAdvanceRound(state: MatchState): MatchState {
+  if (!state.roundComplete) return state;
+  const board = boardScore(state.round);
+  const banked: [number, number] = [state.carry[0] + board[0], state.carry[1] + board[1]];
+  const roundIdx = state.roundIdx + 1;
+  return {
+    ...state,
+    round: resetFromOrder(
+      state.deckOrders[roundIdx],
+      startingPlayer(banked, roundIdx, state.coinFlips),
+    ),
+    roundIdx,
+    carry: banked,
+    roundHistory: [...state.roundHistory, board],
+    roundComplete: false,
+  };
+}

web/src/game/matchObservation.test.ts

@@ -0,0 +1,144 @@
+import { describe, expect, it } from "vitest";
+
+import fixture from "./match-parity-fixture.json";
+import {
+  N_ROUNDS,
+  matchAdvanceRound,
+  matchFromOrders,
+  matchLegalActionMask,
+  matchScore,
+  matchStep,
+  roundScore,
+  startingPlayer,
+  type MatchState,
+} from "./match";
+import { MATCH_OBS_DIM, matchObservation } from "./matchObservation";
+import type { GameState } from "./types";
+
+interface FixtureRow {
+  match: {
+    round: GameState;
+    deckOrders: number[][];
+    coinFlips: number[];
+    roundIdx: number;
+    carry: number[];
+    done: boolean;
+  };
+  player: number;
+  observation: number[];
+}
+
+const rows = fixture.rows as unknown as FixtureRow[];
+
+function toMatch(row: FixtureRow): MatchState {
+  return {
+    round: row.match.round,
+    deckOrders: row.match.deckOrders,
+    coinFlips: row.match.coinFlips,
+    roundIdx: row.match.roundIdx,
+    carry: [row.match.carry[0], row.match.carry[1]],
+    // The fixture is generated from the JAX match, which is always three rounds.
+    totalRounds: N_ROUNDS,
+    // Not part of the observation; only present so the type is satisfied.
+    roundComplete: false,
+    roundHistory: [],
+    done: row.match.done,
+  };
+}
+
+function firstLegal(match: MatchState): number {
+  const index = matchLegalActionMask(match).findIndex(Boolean);
+  if (index < 0) throw new Error("no legal action");
+  return index;
+}
+
+describe("match observation parity with JAX", () => {
+  it("has the dimension the exported model expects", () => {
+    expect(MATCH_OBS_DIM).toBe(fixture.obsDim);
+    expect(MATCH_OBS_DIM).toBe(501);
+  });
+
+  it("reproduces every fixture observation", () => {
+    expect(rows.length).toBeGreaterThan(100);
+    let worst = 0;
+    let worstAt = "";
+
+    for (const [index, row] of rows.entries()) {
+      const actual = matchObservation(toMatch(row), row.player);
+      expect(actual.length).toBe(row.observation.length);
+
+      for (let i = 0; i < actual.length; i += 1) {
+        const delta = Math.abs(actual[i] - row.observation[i]);
+        if (delta > worst) {
+          worst = delta;
+          worstAt = `row ${index}, feature ${i}`;
+        }
+      }
+    }
+    // A mismatch here throws nowhere: the ONNX policy consumes the wrong vector
+    // and plays worse for reasons nobody can see. So the bar is float32 round-off,
+    // not "close enough".
+    expect(worst, `largest disagreement at ${worstAt}`).toBeLessThan(1e-5);
+  });
+
+  it("covers positions past a round roll-over, with a real carry", () => {
+    expect(rows.some((row) => row.match.roundIdx > 0)).toBe(true);
+    expect(rows.some((row) => row.match.carry[0] !== row.match.carry[1])).toBe(true);
+  });
+});
+
+/** Play a full match, advancing past each round-complete pause. */
+function playMatch(start: MatchState): { final: MatchState; pauses: number } {
+  let match = start;
+  let pauses = 0;
+  for (let ply = 0; ply < 1400 && !match.done; ply += 1) {
+    if (match.roundComplete) {
+      pauses += 1;
+      match = matchAdvanceRound(match);
+      continue;
+    }
+    match = matchStep(match, firstLegal(match));
+  }
+  return { final: match, pauses };
+}
+
+describe("match rules", () => {
+  it("pauses at each round end, then plays exactly three rounds", () => {
+    const start = matchFromOrders(rows[0].match.deckOrders, rows[0].match.coinFlips);
+    const { final, pauses } = playMatch(start);
+    // Two roll-overs for three rounds; the third ends the match, not a pause.
+    expect(pauses).toBe(N_ROUNDS - 1);
+    expect(final.done).toBe(true);
+    expect(final.roundIdx).toBe(N_ROUNDS - 1);
+    expect(final.roundHistory.length).toBe(N_ROUNDS);
+  });
+
+  it("holds the finished board while round-complete, offering no legal move", () => {
+    let match = matchFromOrders(rows[0].match.deckOrders, rows[0].match.coinFlips);
+    while (!match.roundComplete && !match.done) match = matchStep(match, firstLegal(match));
+    expect(match.roundComplete).toBe(true);
+    expect(match.roundIdx).toBe(0); // not advanced yet
+    expect(matchLegalActionMask(match).some(Boolean)).toBe(false);
+    // The running total already includes the round being shown.
+    expect(matchScore(match)).toEqual(roundScore(match));
+  });
+
+  it("banks each round exactly once into the history and carry", () => {
+    const { final } = playMatch(matchFromOrders(rows[0].match.deckOrders, rows[0].match.coinFlips));
+    const summed = final.roundHistory.reduce(
+      (acc, [you, rival]) => [acc[0] + you, acc[1] + rival] as [number, number],
+      [0, 0] as [number, number],
+    );
+    // carry banks rounds 1..2; the third is still on the board at match end.
+    const board = roundScore(final);
+    expect(matchScore(final)).toEqual([summed[0], summed[1]]);
+    expect([final.carry[0] + board[0], final.carry[1] + board[1]]).toEqual(matchScore(final));
+  });
+
+  it("lets whoever has more points begin, and flips a coin when level", () => {
+    expect(startingPlayer([60, 10], 1, [1, 1, 1])).toBe(0); // ahead leads, coin ignored
+    expect(startingPlayer([10, 60], 1, [0, 0, 0])).toBe(1);
+    expect(startingPlayer([30, 30], 1, [0, 0, 0])).toBe(0); // level falls back to the coin
+    expect(startingPlayer([30, 30], 1, [1, 1, 1])).toBe(1);
+  });
+});

web/src/game/matchObservation.ts

@@ -0,0 +1,140 @@
+import { cardColor } from "./cards";
+import { boardScore } from "./engine";
+import { N_ROUNDS, type MatchState } from "./match";
+import { observation } from "./observation";
+import {
+  CARDS_PER_COLOR,
+  LOC_DECK,
+  LOC_DISCARD,
+  LOC_P0_HAND,
+  N_CARDS,
+  N_COLORS,
+  OBS_DIM,
+} from "./types";
+
+/**
+ * Player-view observation for a three-round match: the mirror of
+ * src/lost_cities_jax/match_obs.py.
+ *
+ * These two must agree to the bit. A mismatch does not throw -- the model happily
+ * consumes a wrong vector and plays worse for reasons nobody can see. That is why
+ * matchObservation.test.ts checks this against fixtures generated from the Python
+ * side rather than trusting the port.
+ *
+ * On top of the single-round observation this adds the four things a match policy
+ * cannot play without: carry (scalar *and* binned, because round three is a
+ * threshold problem); which round it is; whose turn it is; and the deck clock,
+ * since a round ends on the last deck draw and players bend that parity by drawing
+ * from discard piles. Live points per colour are split by hand / discard pile /
+ * unseen, because a discard pile is public and recoverable.
+ */
+
+/** Packed tightly around zero: that is where the round-three decision flips. */
+const CARRY_BIN_EDGES = [-60, -30, -12, -1, 1, 12, 30, 60];
+export const N_CARRY_BINS = CARRY_BIN_EDGES.length + 1;
+
+/** A typical round margin, not the theoretical 780 maximum the old obs divided by. */
+const CARRY_SCALE = 75;
+
+/** 2+3+...+10, the most one expedition can be worth before multipliers. */
+const MAX_COLOR_POINTS = 54;
+
+const N_MATCH_SCALARS =
+  1 + N_CARRY_BINS + N_ROUNDS + 1 + 1 + 1 + 1 + 2 * N_COLORS * 3;
+export const MATCH_OBS_DIM = OBS_DIM + N_MATCH_SCALARS;
+
+function cardRank(card: number): number {
+  const slot = card % CARDS_PER_COLOR;
+  return slot >= 3 ? slot - 1 : 0;
+}
+
+/** numpy.digitize: the count of edges strictly below `value`. */
+function digitize(value: number, edges: readonly number[]): number {
+  let index = 0;
+  while (index < edges.length && value >= edges[index]) index += 1;
+  return index;
+}
+
+/**
+ * Points still reachable for `subject`, split by where the card sits:
+ * [in hand, in a discard pile, unseen] per colour.
+ *
+ * Seen through `viewer`'s eyes -- a card in the opponent's hand only counts as "in
+ * hand" if it is public, otherwise it is unseen.
+ */
+function livePoints(state: MatchState, viewer: number, subject: number): number[] {
+  const round = state.round;
+  const isMine = subject === viewer;
+  const subjectHandLoc = LOC_P0_HAND + subject;
+  const out = Array.from({ length: N_COLORS }, () => [0, 0, 0]);
+
+  for (let card = 0; card < N_CARDS; card += 1) {
+    const color = cardColor(card);
+    const rank = cardRank(card);
+    // An ascending column can only take cards above its current top.
+    if (rank <= round.colTop[subject][color]) continue;
+
+    const loc = round.cardLoc[card];
+    const inSubjectHand = loc === subjectHandLoc;
+    if (inSubjectHand && (isMine || round.handPublic[card])) {
+      out[color][0] += rank;
+    } else if (loc === LOC_DISCARD) {
+      out[color][1] += rank;
+    } else if (loc === LOC_DECK || (inSubjectHand && !isMine && !round.handPublic[card])) {
+      // Cards in the *other* player's hidden hand are unseen to the viewer too,
+      // but the subject cannot reach them, so they are deliberately excluded.
+      out[color][2] += rank;
+    }
+  }
+  return out.flat().map((points) => points / MAX_COLOR_POINTS);
+}
+
+export function matchObservation(state: MatchState, player: number): Float32Array {
+  const opponent = 1 - player;
+  const round = state.round;
+  const base = observation(round, player);
+
+  const board = boardScore(round);
+  const lead =
+    state.carry[player] + board[player] - (state.carry[opponent] + board[opponent]);
+
+  const carryScaled = Math.min(2, Math.max(-2, lead / CARRY_SCALE));
+  const carryBins = Array.from({ length: N_CARRY_BINS }, (_, i) =>
+    Number(i === digitize(lead, CARRY_BIN_EDGES)),
+  );
+
+  const roundOneHot = Array.from({ length: N_ROUNDS }, (_, i) => Number(i === state.roundIdx));
+  const roundsLeft = (N_ROUNDS - 1 - state.roundIdx) / (N_ROUNDS - 1);
+
+  const toMove = round.toMove;
+  const myTurn = Number(toMove === player);
+  // toMove flips every ply, so the round's opener is recoverable from parity.
+  const roundOpener = toMove ^ (round.stepCount & 1);
+  const iOpened = Number(roundOpener === player);
+
+  // If both players drew from the deck from here, the last deck card falls to
+  // whoever is on move after `remaining - 1` more plies.
+  const remaining = Math.max(N_CARDS - round.drawPtr, 1);
+  const lastDrawer = toMove ^ ((remaining - 1) & 1);
+  const iTakeLast = Number(lastDrawer === player);
+
+  const extra = [
+    carryScaled,
+    ...carryBins,
+    ...roundOneHot,
+    roundsLeft,
+    myTurn,
+    iOpened,
+    iTakeLast,
+    ...livePoints(state, player, player),
+    ...livePoints(state, player, opponent),
+  ];
+
+  if (extra.length !== N_MATCH_SCALARS) {
+    throw new Error(`match scalars ${extra.length} != ${N_MATCH_SCALARS}`);
+  }
+  const out = new Float32Array(MATCH_OBS_DIM);
+  out.set(base, 0);
+  out.set(Float32Array.from(extra), OBS_DIM);
+  return out;
+}

web/src/game/persistence.test.ts

@@ -0,0 +1,52 @@
+import { describe, expect, it } from "vitest";
+
+import { matchFromOrders } from "./match";
+import { matchFromSeed } from "./random";
+import { parseSavedGame } from "./persistence";
+
+const { deckOrders, coinFlips } = matchFromSeed("abc");
+const match = matchFromOrders(deckOrders, coinFlips, 3);
+
+function savedGame(overrides: Record<string, unknown> = {}) {
+  return {
+    version: 2,
+    seed: "abc",
+    mode: 3,
+    frames: [{ state: match, selection: { handSlot: null, placeType: null } }],
+    cursor: 0,
+    resultOpen: true,
+    ...overrides,
+  };
+}
+
+describe("saved game parsing", () => {
+  it("accepts a complete saved timeline", () => {
+    const saved = savedGame();
+    expect(parseSavedGame(JSON.stringify(saved))).toEqual(saved);
+  });
+
+  it("accepts a one-deal game", () => {
+    const saved = savedGame({
+      mode: 1,
+      frames: [
+        { state: matchFromOrders(deckOrders, coinFlips, 1), selection: { handSlot: null, placeType: null } },
+      ],
+    });
+    expect(parseSavedGame(JSON.stringify(saved))).toEqual(saved);
+  });
+
+  it("rejects corrupt and incompatible data", () => {
+    expect(parseSavedGame("not json")).toBeNull();
+    expect(parseSavedGame(JSON.stringify({ version: 2 }))).toBeNull();
+    expect(parseSavedGame(JSON.stringify(savedGame({ frames: [] })))).toBeNull();
+    expect(parseSavedGame(JSON.stringify(savedGame({ mode: 2 })))).toBeNull();
+  });
+
+  it("refuses a v1 save rather than guessing what it meant", () => {
+    // A v1 save is one round: no carry, no deals for rounds two and three, no
+    // coin flips. There is nothing honest to migrate it into, so it is dropped
+    // and a fresh game is dealt.
+    const v1 = { version: 1, seed: "abc", frames: [{ state: match.round, selection: {} }], cursor: 0, resultOpen: true };
+    expect(parseSavedGame(JSON.stringify(v1))).toBeNull();
+  });
+});

web/src/game/persistence.ts

@@ -0,0 +1,99 @@
+import type { MatchState } from "./match";
+import { N_CARDS, type GameState, type PlaceType } from "./types";
+
+/**
+ * v2 because a saved game now holds a match, not a round. The key is bumped
+ * rather than migrated: a v1 save has no carry, no deals for rounds two and
+ * three, and no coin flips, so there is nothing honest to migrate it into. An
+ * unreadable save just deals a fresh game, which is the right failure.
+ */
+export const SAVED_GAME_KEY = "lost-cities-jax-ppo.game.v2";
+
+export interface SavedFrame {
+  state: MatchState;
+  selection: { handSlot: number | null; placeType: PlaceType | null };
+}
+
+export interface SavedGame {
+  version: 2;
+  seed: string;
+  /** 1 for a one-off deal, 3 for the classic match. */
+  mode: number;
+  frames: SavedFrame[];
+  cursor: number;
+  resultOpen: boolean;
+}
+
+function isNumberArray(value: unknown, length?: number): value is number[] {
+  return Array.isArray(value) && (length === undefined || value.length === length) &&
+    value.every((item) => typeof item === "number" && Number.isFinite(item));
+}
+
+function isBooleanArray(value: unknown, length: number): value is boolean[] {
+  return Array.isArray(value) && value.length === length &&
+    value.every((item) => typeof item === "boolean");
+}
+
+function isMatrix(value: unknown, rows: number, columns: number): value is number[][] {
+  return Array.isArray(value) && value.length === rows &&
+    value.every((row) => isNumberArray(row, columns));
+}
+
+function isDeckOrder(value: unknown): value is number[] {
+  return isNumberArray(value, N_CARDS) && new Set(value).size === N_CARDS;
+}
+
+function isGameState(value: unknown): value is GameState {
+  if (typeof value !== "object" || value === null) return false;
+  const state = value as Partial<GameState>;
+  return isDeckOrder(state.deckOrder) &&
+    isNumberArray(state.cardLoc, N_CARDS) && isBooleanArray(state.handPublic, N_CARDS) &&
+    isMatrix(state.colTop, 2, 5) && isMatrix(state.colHandshakes, 2, 5) &&
+    isMatrix(state.colLength, 2, 5) && Array.isArray(state.piles) && state.piles.length === 5 &&
+    state.piles.every((pile) => isNumberArray(pile)) &&
+    typeof state.drawPtr === "number" && typeof state.stepCount === "number" &&
+    (state.toMove === 0 || state.toMove === 1) && typeof state.done === "boolean";
+}
+
+function isMatchState(value: unknown): value is MatchState {
+  if (typeof value !== "object" || value === null) return false;
+  const match = value as Partial<MatchState>;
+  return isGameState(match.round) &&
+    Array.isArray(match.deckOrders) && match.deckOrders.length === 3 &&
+    match.deckOrders.every(isDeckOrder) &&
+    isNumberArray(match.coinFlips, 3) &&
+    Number.isInteger(match.roundIdx) && match.roundIdx! >= 0 && match.roundIdx! < 3 &&
+    isNumberArray(match.carry, 2) &&
+    (match.totalRounds === 1 || match.totalRounds === 3) &&
+    typeof match.roundComplete === "boolean" &&
+    Array.isArray(match.roundHistory) &&
+    match.roundHistory.every((entry) => isNumberArray(entry, 2)) &&
+    typeof match.done === "boolean";
+}
+
+function isSavedGame(value: unknown): value is SavedGame {
+  if (typeof value !== "object" || value === null) return false;
+  const saved = value as Partial<SavedGame>;
+  if (saved.version !== 2 || typeof saved.seed !== "string" ||
+      (saved.mode !== 1 && saved.mode !== 3) ||
+      !Array.isArray(saved.frames) || saved.frames.length === 0 ||
+      !Number.isInteger(saved.cursor) || saved.cursor! < 0 || saved.cursor! >= saved.frames.length ||
+      typeof saved.resultOpen !== "boolean") return false;
+  return saved.frames.every((frame) => {
+    if (typeof frame !== "object" || frame === null || !isMatchState(frame.state)) return false;
+    const selection = frame.selection;
+    return typeof selection === "object" && selection !== null &&
+      (selection.handSlot === null || Number.isInteger(selection.handSlot)) &&
+      (selection.placeType === null || selection.placeType === 0 || selection.placeType === 1);
+  });
+}
+
+export function parseSavedGame(raw: string | null): SavedGame | null {
+  if (raw === null) return null;
+  try {
+    const value: unknown = JSON.parse(raw);
+    return isSavedGame(value) ? value : null;
+  } catch {
+    return null;
+  }
+}

web/src/game/random.ts

@@ -25,6 +25,21 @@ export function deckOrderFromSeed(seed: string): number[] {
   return shuffledDeck(mulberry32(hashSeed(seed)));
 }
 
+/**
+ * Every deal and coin flip of a match, drawn up front from one seed.
+ *
+ * Drawing them all now — rather than shuffling again when a round rolls over —
+ * is what keeps a seed reproducible: the whole match is a pure function of it.
+ * The first deal is `deckOrderFromSeed`, so a match and a one-off deal on the
+ * same seed open on the same position.
+ */
+export function matchFromSeed(seed: string): { deckOrders: number[][]; coinFlips: number[] } {
+  const random = mulberry32(hashSeed(seed));
+  const deckOrders = [shuffledDeck(random), shuffledDeck(random), shuffledDeck(random)];
+  const coinFlips = [0, 1, 2].map(() => (random() < 0.5 ? 0 : 1));
+  return { deckOrders, coinFlips };
+}
+
 /** A fresh shareable seed, e.g. "k3f9qa". */
 export function randomSeed(): string {
   return Math.floor(Math.random() * 36 ** 6).toString(36).padStart(6, "0");

web/src/model/policy.ts

@@ -1,13 +1,25 @@
 import * as ort from "onnxruntime-web/all";
 
-import { currentHandSorted, legalActionMask } from "../game/engine";
-import { observation } from "../game/observation";
-import { DISCARD, DRAW_DECK, PLAY, type GameState } from "../game/types";
+import { currentHandSorted } from "../game/engine";
+import { matchLegalActionMask, type MatchState } from "../game/match";
+import { matchObservation } from "../game/matchObservation";
+import { DISCARD, DRAW_DECK, PLAY } from "../game/types";
 import { cardColor, cardRank, isHandshake } from "../game/cards";
 
 export type ExecutionProvider = "webgpu" | "wasm" | "heuristic";
 
-const MODEL_URL = `${import.meta.env.BASE_URL}models/jax-ppo.onnx`;
+/**
+ * borealis -- see data/models.json. Trained on the three-round match, so it takes
+ * the match view (carry, round, deck clock) rather than a bare round. A one-off
+ * deal is simply its round one at a carry of zero, which is a position it has seen
+ * a great many times.
+ */
+const MODEL_URL = `${import.meta.env.BASE_URL}models/borealis.onnx`;
+
+/** Identity of what actually plays. Records carry the hash; the codename is for
+ *  humans and is assigned in data/models.json, not derived. */
+export const MODEL_CODENAME = "borealis";
+export const MODEL_HASH = "13a25243de1c";
 
 /** A legal action with the policy's confidence in it, if the policy has one. */
 export interface RankedAction {
@@ -18,16 +30,16 @@ export interface RankedAction {
 export interface Policy {
   readonly provider: ExecutionProvider;
   /** Legal actions, best first. Drives both the rival's move and the hint. */
-  rank(state: GameState): Promise<RankedAction[]>;
-  action(state: GameState): Promise<number>;
+  rank(match: MatchState): Promise<RankedAction[]>;
+  action(match: MatchState): Promise<number>;
 }
 
 abstract class RankingPolicy implements Policy {
   abstract readonly provider: ExecutionProvider;
-  abstract rank(state: GameState): Promise<RankedAction[]>;
+  abstract rank(match: MatchState): Promise<RankedAction[]>;
 
-  async action(state: GameState): Promise<number> {
-    const [best] = await this.rank(state);
+  async action(match: MatchState): Promise<number> {
+    const [best] = await this.rank(match);
     if (best === undefined) throw new Error("state has no legal actions");
     return best.action;
   }
@@ -54,12 +66,12 @@ class OnnxPolicy extends RankingPolicy {
     super();
   }
 
-  async rank(state: GameState): Promise<RankedAction[]> {
-    const obs = observation(state, state.toMove);
+  async rank(match: MatchState): Promise<RankedAction[]> {
+    const obs = matchObservation(match, match.round.toMove);
     const result = await this.session.run({ obs: new ort.Tensor("float32", obs, [1, obs.length]) });
     const logits = result.logits?.data;
     if (!logits) throw new Error("ONNX model did not return a logits output");
-    const legal = legalActionMask(state);
+    const legal = matchLegalActionMask(match);
     return softmaxOverLegal(legal.map((_, action) => Number(logits[action])), legal);
   }
 }
@@ -67,8 +79,9 @@ class OnnxPolicy extends RankingPolicy {
 export class HeuristicPolicy extends RankingPolicy {
   readonly provider = "heuristic" as const;
 
-  async rank(state: GameState): Promise<RankedAction[]> {
-    const legal = legalActionMask(state);
+  async rank(match: MatchState): Promise<RankedAction[]> {
+    const state = match.round;
+    const legal = matchLegalActionMask(match);
     const hand = currentHandSorted(state);
     const ranked = legal.flatMap((isLegal, action) => {
       if (!isLegal) return [];

web/src/styles.css

@@ -18,6 +18,23 @@
 }
 
 * { box-sizing: border-box; }
+
+/* The table is a control surface, not a document. Dragging across it should never
+   paint a card's number blue or pop a long-press copy menu on mobile. Selection is
+   off everywhere and turned back on only for text you would actually want to copy
+   or edit -- in practice the seed field. user-select does not touch click or focus,
+   so every button still works. */
+.app-shell {
+  user-select: none;
+  -webkit-user-select: none;
+  -webkit-touch-callout: none;
+}
+
+.app-shell input,
+.app-shell textarea {
+  user-select: text;
+  -webkit-user-select: text;
+}
 html, body, #root { min-width: 1024px; min-height: 100%; margin: 0; }
 /* Below the supported minimum width/height the board no longer shrinks
    (app-shell keeps a 1024x720 floor), so allow scrolling instead of clipping
@@ -108,7 +125,7 @@ button:focus-visible { outline: 2px solid var(--gold); outline-offset: 4px; }
   display: flex;
   transform: translateX(-50%);
 }
-.opponent-hand .card { margin-left: -17px; }
+.opponent-hand .card { margin-left: -14px; }
 .opponent-hand .card:first-child { margin-left: 0; }
 
 .menu-wrap { position: absolute; top: 16px; right: 24px; z-index: 20; }
@@ -316,28 +333,10 @@ button.lane__discard:disabled { opacity: 1; }
   pointer-events: none;
 }
 
-/* The hinted card, its destination, and its draw source. */
-.card--hinted { box-shadow: 0 0 0 3px var(--gold), 0 0 26px rgba(201, 163, 75, 0.35); }
-.lane__zone.is-hinted,
-.lane__discard.is-hinted,
-.deck-stack.is-hinted { animation: target-pulse 1.6s ease-in-out infinite; }
-.lane__zone.is-hinted .lane__ghost,
-.lane__discard.is-hinted .lane__ghost {
-  border-color: var(--gold);
-  border-style: dashed;
-  opacity: 0.9;
-}
-.lane__zone.is-hinted .card,
-.lane__discard.is-hinted .card,
-.deck-stack.is-hinted .card { box-shadow: 0 0 0 2px var(--gold), 0 0 26px rgba(201, 163, 75, 0.3); }
-
 @media (prefers-reduced-motion: reduce) {
   .lane__zone.is-target,
   .lane__discard.is-target,
   .lane__discard.is-draw-target,
-  .lane__zone.is-hinted,
-  .lane__discard.is-hinted,
-  .deck-stack.is-hinted,
   .deck-stack.is-draw-target { animation: none; }
 }
 
@@ -367,15 +366,28 @@ button.card:disabled { cursor: default; }
 }
 .card__corner {
   position: absolute;
-  left: 8px;
+  left: 6px;
   display: flex;
   flex-direction: column;
   align-items: center;
   color: var(--card-color);
   font-family: Georgia, serif;
 }
-.card__corner--top { top: 7px; }
+/* Cards overlap left-over-right, so the only strip that stays visible is each
+   card's left edge -- exactly where the top corner sits. It carries a small ivory
+   backing so the index reads even when it falls over the card behind it, which is
+   what lets the hand overlap tightly and still be legible. */
+.card__corner--top {
+  top: 6px;
+  z-index: 2;
+  padding: 2px 5px 3px;
+  border-radius: 7px;
+  background: color-mix(in srgb, var(--ivory) 82%, transparent);
+  box-shadow: 0 1px 3px rgba(0, 0, 0, 0.28);
+}
 .card__corner--bottom { right: 8px; bottom: 7px; left: auto; }
+.card__corner--top b { font-size: 20px; font-weight: 700; }
+.card__corner--top i { margin-top: 0; font-size: 13px; }
 .card__corner b { font-size: 18px; line-height: 1; }
 .card__corner i { margin-top: 1px; font-size: 12px; font-style: normal; line-height: 1; }
 .card__center {
@@ -418,11 +430,25 @@ button.card:disabled { cursor: default; }
   border-width: 2px;
   border-radius: 5px;
 }
-.card--mini .card__corner { display: none; }
-.card--mini .card__center { inset: 4px; }
-.card--mini .card__center b { font-size: 20px; }
-.card--mini .card__center i { margin-top: 2px; font-size: 11px; }
-.card--mini .card__center--handshake i { font-size: 19px; }
+/* The rival's revealed cards overlap -14px, leaving a ~30px strip. The centered
+   value is buried on every card but the top one, so a small top-left index rides
+   in the visible strip -- the same trick as the player's hand, scaled down. */
+.card--mini .card__corner--bottom { display: none; }
+.card--mini .card__corner--top {
+  top: 3px;
+  left: 3px;
+  padding: 1px 3px 2px;
+  border-radius: 5px;
+  flex-direction: row;
+  gap: 2px;
+  align-items: baseline;
+}
+.card--mini .card__corner--top b { font-size: 13px; }
+.card--mini .card__corner--top i { font-size: 8px; }
+.card--mini .card__center { inset: 4px; top: 14px; }
+.card--mini .card__center b { font-size: 17px; }
+.card--mini .card__center i { margin-top: 2px; font-size: 10px; }
+.card--mini .card__center--handshake i { font-size: 16px; }
 .card--mini .card__center small { display: none; }
 .card--mini::after { inset: 3px; border-radius: 3px; }
 .card--mini .card__flip-cover { border-radius: 5px; }
@@ -472,7 +498,7 @@ button.card:disabled { cursor: default; }
 .deck-stack > span { margin-top: 5px; color: #7e817c; font: 700 9px Arial, sans-serif; letter-spacing: 0.2em; }
 .deck-stack.is-draw-target .card { box-shadow: 0 0 0 3px var(--gold), 7px 8px 0 #171b1e, 0 0 32px rgba(201, 163, 75, 0.3); }
 
-/* Insets keep the row clear of the history bar parked at the right. */
+/* Insets keep the prompt clear of the controls parked at the right. */
 .prompt-row {
   position: absolute;
   right: 330px;
@@ -497,20 +523,6 @@ button.card:disabled { cursor: default; }
 .turn-prompt--thinking { color: #c6b887; }
 .turn-prompt--review { color: #9fb4c8; }
 
-.hint-button {
-  flex: 0 0 auto;
-  height: 32px;
-  padding: 0 14px;
-  border: 1px solid var(--gold-dim);
-  border-radius: 9px;
-  background: rgba(7, 13, 12, 0.9);
-  font: 700 9px Arial, sans-serif;
-  letter-spacing: 0.16em;
-  cursor: pointer;
-}
-.hint-button:hover:not(:disabled) { border-color: var(--gold); }
-.hint-button:disabled { opacity: 0.35; cursor: default; }
-.hint-button.is-active { border-color: var(--gold); color: var(--gold); }
 
 .human-hand {
   position: absolute;
@@ -610,23 +622,25 @@ button.card:disabled { cursor: default; }
 .result-card button:hover { border-color: var(--gold); }
 .result-card button.is-primary { border-color: var(--gold); color: var(--gold); }
 
-/* Sits in the band above the hand: the hand row is centered and its width is
-   computed against the score plaques, so a wide bar down at the hand's level
-   would overlap the leftmost/rightmost cards. */
-.history-bar {
+/* Player controls share the hand row's bottom edge and stay in the mirrored
+   right-side keep-out reserved by the centered hand layout. */
+.control-stack {
   position: absolute;
   right: 24px;
-  bottom: 196px;
+  bottom: 30px;
   z-index: 12;
   display: flex;
-  align-items: center;
+  width: 142px;
+  flex-direction: column;
+  align-items: stretch;
   gap: 8px;
 }
-.history-bar button {
+.control-stack button {
   height: 32px;
   padding: 0 11px;
   display: flex;
   align-items: center;
+  justify-content: center;
   gap: 6px;
   border: 1px solid var(--gold-dim);
   border-radius: 9px;
@@ -634,15 +648,11 @@ button.card:disabled { cursor: default; }
   font: 700 13px/1 Georgia, serif;
   cursor: pointer;
 }
-.history-bar button span { font: 700 9px Arial, sans-serif; letter-spacing: 0.14em; }
-.history-bar button:hover:not(:disabled) { border-color: var(--gold); }
-.history-bar button:disabled { opacity: 0.3; cursor: default; }
-.history-bar__resume {
-  border-color: var(--gold) !important;
-  color: var(--gold);
-  font: 700 9px Arial, sans-serif;
-  letter-spacing: 0.14em;
-}
+.control-stack button span { font: 700 9px Arial, sans-serif; letter-spacing: 0.14em; }
+.control-stack button:hover:not(:disabled) { border-color: var(--gold); }
+.control-stack button:disabled { opacity: 0.3; cursor: default; }
+.control-stack__score { visibility: hidden; }
+.control-stack__score.is-available { visibility: visible; }
 
 @media (max-width: 1250px) {
   .table-center { width: 600px; }
@@ -651,7 +661,6 @@ button.card:disabled { cursor: default; }
   .human-hand .card { width: 98px; height: 142px; margin-left: 7px; }
   .human-hand { height: 146px; }
   .prompt-row { bottom: 177px; }
-  .history-bar { bottom: 172px; }
   .table-center { bottom: 210px; }
   /* Narrow viewports: the plaques are the keep-out that squeezes the hand row,
      so they become compact chips (model-name line dropped) to give the cards
@@ -675,7 +684,96 @@ button.card:disabled { cursor: default; }
   .human-hand { bottom: 18px; height: 138px; }
   .human-hand .card { width: 94px; height: 134px; }
   .prompt-row { bottom: 165px; }
-  .history-bar { bottom: 160px; }
+  .control-stack { bottom: 18px; }
   .score-plaque--human { bottom: 9px; }
   .deck-stack { transform: scale(0.86); }
 }
+
+/* Match mode: which round is live, and what is already banked. The board only
+   ever shows the round in play, so without this the score plaques would be the
+   only hint that two more rounds are coming. */
+/* Tucked under the rival plaque on the left, clear of the opponent hand which
+   sits centered along the top and would otherwise cover it. */
+.round-strip {
+  position: absolute;
+  top: 88px;
+  left: 24px;
+  display: flex;
+  align-items: center;
+  gap: 8px;
+  padding: 6px 14px;
+  border-radius: 999px;
+  background: rgba(18, 14, 10, 0.72);
+  border: 1px solid rgba(214, 188, 140, 0.22);
+  font-size: 11px;
+  letter-spacing: 0.12em;
+  color: rgba(214, 188, 140, 0.55);
+  z-index: 9;
+  pointer-events: none;
+}
+
+.round-strip span {
+  font-weight: 700;
+}
+
+.round-strip span.is-done {
+  color: rgba(214, 188, 140, 0.85);
+}
+
+.round-strip span.is-live {
+  color: #f0d9a6;
+  text-shadow: 0 0 10px rgba(240, 217, 166, 0.45);
+}
+
+.round-strip em {
+  font-style: normal;
+  margin-left: 4px;
+  padding-left: 10px;
+  border-left: 1px solid rgba(214, 188, 140, 0.22);
+  color: rgba(214, 188, 140, 0.75);
+}
+
+.menu-modes {
+  display: flex;
+  gap: 6px;
+}
+
+.menu-modes button {
+  flex: 1;
+  font-size: 10px;
+}
+
+.menu-modes button.is-active {
+  background: rgba(240, 217, 166, 0.16);
+  color: #f0d9a6;
+  border-color: rgba(240, 217, 166, 0.45);
+}
+
+/* A match's headline is the summed total; the table still breaks down the round
+   in front of you, so the two have to be told apart. */
+.result-card__banked {
+  display: block;
+  margin-top: 6px;
+  font-size: 11px;
+  letter-spacing: 0.1em;
+  color: rgba(214, 188, 140, 0.55);
+}
+
+.result-table__match td,
+.result-table__match th {
+  border-top: 1px solid rgba(214, 188, 140, 0.22);
+  color: #f0d9a6;
+  font-weight: 700;
+}
+
+/* The undo/redo settings switch. Off by default: taking moves back is a training
+   aid, and the honest game is the sensible default. */
+.menu-toggle {
+  display: flex;
+  align-items: center;
+  gap: 8px;
+  padding: 6px 4px;
+  cursor: pointer;
+}
+.menu-toggle input { width: 15px; height: 15px; accent-color: var(--gold); cursor: pointer; }
+.menu-toggle span { padding: 0; color: #9a9d97; }

web/src/ui/useCardMotion.ts

@@ -153,11 +153,22 @@ export function useCardMotion(): CardMotion {
         continue;
       }
 
+      // getBoundingClientRect includes a transform applied by WAAPI. A render
+      // while a deal/move is still in flight (for example when the policy
+      // finishes loading) must retain the intended destination instead of
+      // treating the animated visual position as a new layout and launching a
+      // second flight.
+      const previous = rects.current.get(card);
+      if (element.getAnimations().some((animation) => animation.playState === "running")) {
+        if (previous) next.set(card, previous);
+        continue;
+      }
+
       const to = element.getBoundingClientRect();
       next.set(card, to);
       if (reduced) continue;
 
-      const from = rects.current.get(card);
+      const from = previous;
       if (from) {
         animateMove(element, from, to);
       } else if (deckRect) {