- Shared card pool and item stock as in the official mode (the owner's request, from the 路标档案馆 video and public notes): solo low-tier exceptions, co-op operators x2, shared item stock with special-item exceptions, normal / advanced occupancy 1 / 2, re-check on buy, 机变 exhaustion, release on consume / destroy / elimination; 整备 never exceeds stock; invariant held <= cap.
- 突袭: one taunt-first reselect right after a revive (the owner's rule); ordinary polls and redeploys keep the 0.2.3 rule; a revive on an enemy's tile stays. #475: online boss / hidden-boss Raid may land on the empty half (owner-limited scope).
- 浊心斯卡蒂 S1 x 坚守 no longer bounce redirected damage, and the two shares conserve the hit (idea of PR #464 by @HappyTanuki).
- Operator fixes adapted from PRs by @Xinglan233 (#459 哈洛德 S2 elemental trigger, #460 涤火杰西卡 shield on the four adjacent tiles, #461 家族徽章 first hit after stealth, #462 松果 S2 regrow per deploy); 克莱门莎 S2 splash 1.1 / 100% ATK and air auto-trigger (#467); 白面 copying 歌蕾蒂娅 / 余 skips the three-member condition (#473, the owner's video).
- 香槟炸弹 is not chosen by friendly auras / heals (PR #468 / #476 by @Siornya); 琳琅诗怀雅 coin balance (PR #450 by @IceCodeNew); Touch prep display (PR #451 by @YinqiMargatroid); lobby header (PR #471 by @YuriRestia); Japanese UI terms (PR #448 by @scutkane).
- AI-only boss countdown grace, render-failure cache / recovery without restarting a battle, device units kept on a same-field re-entry, DIY detail / skill previews, large-text result buttons.
- golden: battle families equal 0.2.3 except diy-135 / diy-136 (克莱门莎); all 18 release matches move through the shared pool, every flip named in docs/history/0.2.4.md §29.12; two coverage scenarios added.
Guard lethal stuns against a replacement deployment; keep bard auras running
through stun; require an active Aegir bond for Clementia's strategy; allow
Compass SP to bypass recovery locks; and refresh Hidden Core gun mitigation
against current gun/spring companions. Set multiplayer strategy picks to 50s
and describe renderer fallback without diagnosing unsupported hardware.
Golden moves (full 285-scenario regeneration):
- roster-008: inactive Aegir no longer gains 18 strategy layers.
- boss-boss_4-solo: bard regeneration continues through stun.
- hidden-boss_9-pair: mitigation covers first hits/current companions.
- standin-07: hammer stun cannot follow an immediate revival deployment.
No ending-reason changes. Sources and report-by-report findings: DESIGN 28.25.
docs/SIM.md §2: the countdown (attackCountdown, 1e-9 like timeLeft / buffs / every()), 1 s = 30 ticks and 3 s = 90, the
双眼皮 turrets and 寒檀's icicles on it, the drones that compared with 1e-9 already, the count-up content clocks not
covered; the official note — PRTS 作战机制/sandbox 「帧对齐机制对攻速的影响」: one frame = 1/30 s, since 2019-12-24 a
non-integer frame count is rounded (四舍五入; before, 进一取整), an exact one stays; here a non-integer count still ends on
the tick that crosses 0 (⌈30 × interval⌉: 1.25 s = 38, 能天使's 0.678 s 21 against the official 20) — an older
difference, unchanged. §7.1: SP within 1e-9 of the cost is the cost (gainSp, _normalize). design/geometry-time.md §4:
the engine's timer tolerance in one place. history/0.2.2.md §27.61 (Seen / Official / Now / Golden with the end-result
flips one line each / Tests / Docs) and the area map's "→ §27.61". docs-consistency + docs-paths: 31 pass.
- The owner asked to verify the official rule before deciding (research: PRTS 失衡位移机制 / 推与拉 / 游戏数据基础 / 特殊机制,
the client's Rigidbody2D data); the PR's 0.6387·√tiles formula and fear exemption were not taken.
- Grok review: MERGE AFTER FIXES — two HIGH (static pulls, 弧光锋卫's bleed), skills during the hold, husk 重生, exact
frames — fixed on the branch (1db84d79).
- Golden union: bond-miraShip-low differs from HEAD (fewer enemy attacks: 169 → 166 events, same kills) though the
branch alone left it — HEAD's ×1.1 练度 path now displaces an enemy there. Broad node --test 5681 (5679 pass, 0 fail,
2 skipped); lint 0 errors.
Branch review of fb7-unbalance (MERGE AFTER FIXES):
1. A 静态刚体 pulled got the 0.1 s floor: _staticForce now knows push from pull — a push's force is instant (推与拉), so
0.1 s; a pull's lasts its window (pullUnbalance: 1 s, 0.5 s below −1) — PRTS 特殊机制 静态刚体 「在失去施力后立刻」 out,
「0.1 秒保底」. pull() checks the puller's own block before the static case.
2. 弧光锋卫 (PRTS 天赋 「处于失衡状态时，每0.066s受到400点无来源真实持续伤害」, data unbalanced_bleed): a tick bleed of 400 every
0.066 s while unbalancedNow holds; the per-tile one-shot from the position jump ([ASSUMED], UNBALANCE_SPEED) is gone.
A 受力等级 0 push (0.8 s) → 12 hits = 4800 (was 2037), a 急停 pull with no movement (1 s) → 15 = 6000 (was 0), a
wall-shortened slide keeps the 0.8 s row → 4800.
3. 失衡免疫 「失衡期间无法自主移动、发动攻击、使用技能」: enemies/helpers.js unbalancedNow; canCast(e, sil, battle) refuses while it
runs (content skills; the boss parts' calls pass the battle), blinkForward returns null, 枯朽之种's dive / blast and the
铳 / 碎铳之簧 charges wait, 乌顶巨角卢鲁's clash waits for its end. A route APPEAR relocation clears it (a forced state
switch, 失衡位移机制). Left running on purpose (docs/SIM.md): 水遁忍者's damage check (异常效果 「仍然能正常造成伤害」), passive
auras and terrain / periodic effects, death / damage-taken triggers, skill cooldowns, 弧光锋卫's bleed itself.
4. husk() (逐火 余烬 重生) clears the state like reborn().
5. The 35/30 row held 36 frames (t0 + 35/30 overshoots in floats): every check compares with 1e-9; the frames are asserted.
Tests: test/sim/unbalance.test.js +6 (exact frames 6 / 12 / 24 / 27 / 32 / 35 with the walk on the next frame; static pull
1 s / 0.5 s, push 0.1 s; 弧光锋卫 4800 / 6000 / 4800; canCast / blinkForward refused, 枯朽之种 held then diving; the elk's clash
after the state; APPEAR and a 逐火 重生 clear it — all six fail on the parent); enemies_bosses (失衡 kit test) and
feedback1d-solvent (BUFF damage) updated to the state bleed. test/sim + test/content + test/render pass. Docs: SIM.md
(push / pull row), PLAYING.md, design/engine.md (helpers, enemy AI), constants / displacement comments.
Golden against the previous commit — outcome moves: roster-010 (歌蕾蒂娅 pulls 妖怪 drones, now held 1 s: killed 44→45,
leaks 58→57), roster-040 (static pulls of 妖怪: killed 28→25, leaks 37→40, deaths 29→30), roster-043 / 044 (山: deaths 14→16 /
14→13), hidden-boss_9-pair (锏: boss HP 3636991→3586929), coop4-ABYSS-7 (ai_2 LP +1 from round 5), diy-115 (黑键: leaks
70→75), diy-116 (黑键: killed 2→3, leaks 121→120) — the static pulls (probe) and the 1e-9 end (a frame shorter where the float
sum overshot: the 35/30 row, some pull windows); other moves (damage / snapshots only): roster-008 / 017 / 039, 9 bonds,
boss-boss_4-pair / boss_6-pair / hidden-boss_10-pair, coop2-NORMAL-14-standins / coop2-NORMAL-3 / solo-ABYSS-1 /
solo-HARD-1 / solo-HARD-9-boosted / solo-NORMAL-1, diy-053 / 085 / 102.
- Conflicts (the branch started at ae22335c): the four i18n packs (union of keys), game.js DetailPanel props (ops + the
voices' voice=true), detailPanel.js resolveDetail (cultOpts + the voices' tap), match/views.js (both helpers),
docs/PLAYING.md and design/network.md merged line by line against the base. test/match/bosshp.test.js's ATK formula
now includes the unit's 练度 multiplier (cultMul).
- Grok review: MERGE AFTER FIXES — the note and the module card at the set potential fixed on the branch.
- Golden: every owned operator ×1.1 ATK / DEF / max HP by default — roster 49, bonds 46, fields 22, matches 18, diy 134
(stand-in battles and prototype picks unchanged). Broad node --test 5666 (0 fail after the bosshp formula; 2 skipped);
lint 0 errors; typecheck; i18n strict clean (1152 msgids).
- Conflicts (the branch started at ae22335c): spec.js / protocol.js b.progress (resolved from fb7-ui-backlog + leaksBy),
gameLogic.js / game.js exports (device tips + normalizePersonalChoice), docs/PLAYING.md / design/network.md lines
merged against the base, META.md (画卷 note + 教鞭), test/docs-consistency.test.js pin (both fields).
- Grok review: MERGE AFTER FIXES — the HIGH (a forged boss result paid the cards) fixed on the branch (66a228c6), the
META wording (1ab12763).
- Golden: matches 10 (the 杜宾 bot picks and the guard placement); union: coop3-ABYSS-5 differs from HEAD only in ai_2's
dmgDealt (1238011 → 1186269), no outcome change — HEAD's side had moved it, the new placement applies on that path.
- Broad node --test 5623 (5621 pass, 0 fail, 2 skipped); lint 0 errors; typecheck; i18n strict clean.
Seen (review of this branch): the §2.5 bullet said an engine-played seat, "one that left" among them, picks like the bots do,
and design/match.md called the finalizer a "disconnected-seat" one. onLeave -> _quit -> PlayerState.eliminate clears the
bounties and personalChoice (META §1.3 already says the offer goes with the seat); a merely disconnected human is not bot
controlled — it waits for the prep's deadline like for any prep action. The sentences now say that. No code change.
Seen (review of this branch): the Final Assault / Hidden Core path trusts a client's b.result — validateClientResult skips
the leak multiset for boss fields and nothing re-simulates — while the team LP a field costs arrives through b.progress.leaks.
Commit 05eb2b80 paid the perfect-payout cards from the result's leaked list, so a modified authority could have the server
charge a leak and then report two empty lists (LP 58 -> 57 charged, both seats +2: FORGE_BOTH_EMPTY) or pin the leak on the
partner's list (the forger +2, the partner 0: FORGE_PAY).
Now: b.progress carries `leaksBy` (shared/protocol.js, optional; spec.js battleProgress for a boss / hidden field): per
player, the LP of the leak entries the result will show (compactResult's list, lpr clamped as on the wire); `leaks` minus its
sum is the leader's own "扣除目标生命" effects (boss "Doom" / 斥退 / 萨米 Doom), which are no leak and must not cost a perfect
seat its card. The browser runner and the test SimClient send it; the server keeps the highest value per player (f.leaksBy).
Match._bossLeaksAgree (called by _finishFinal) pays a seat of a client-reported field only when (a) the LP the authority
reported was split at all (leaksBy seen whenever lpReported > 0 — a report without the split explains nothing) and (b) the
seat's leaked list shows at least the leak LP reported for it. A server-run field (no client, a takeover) is the truth itself.
Not stopped, and said so in the docs: an authority that lies consistently (a split that blames the partner, a leak labelled as
the leader's effect) — the boss path has no second source; what is caught is a result that contradicts its own progress.
Tests (test first; the three forge tests failed on the parent with [2, 2] / [2, 0] paid, the runner test on the missing field):
clientCombat.test.js — charged leak + two empty lists -> 0 / 0; the leak moved to the partner -> 0 / 0 (with and without a
split); honest split -> the clean seat paid, the leaker not, a leader's effect LP costs nobody; two honest SimClients end to
end; the protocol field; runner.test.js — the browser runner's boss progress carries leaksBy; docs-consistency pins the
b.progress signature. Grok's two repro scripts now print +0 / +0.
Golden: unchanged (matches 18 / full 283 scenarios match; every golden boss field is server-run).
The owner's decisions of 2026-10-08: 「D12 [PR #301] 可以做，具体布局怎么优化你来决定，他那样确实方便，但是三个竖排有点乱影响
观感」 and 「D7D8调配干员里自己设置吧，默认满潜满加成」. The quick choices are PR #301's idea (by @farridge); the layout is new:
one list instead of three columns of cards.
- screens/loadout.js: RosterHead + RosterRow — every row the same four columns under a sticky head (干员 | 技能 |
模组 精锐 | 潜能 · 练度): portrait / name / bonds (selects, opens the detail), three skill slots (an empty one for a
two-skill chess; the label says slot, name, SP, duration — quickSkillTags), the elite's modules (不装备 first), the
潜能 / 练度 selects; a quick choice goes through setChoice / setEntries like the detail and keeps a phone on the list.
The detail's 普通 / 精锐 skill toggle is one preview for the skills, the 特质 and the rows. The detail opens with the
潜能与练度 section; 恢复默认 / 全部恢复默认 reset the settings too (全部: the roster's operators — a 自选 pick's stay
with the 自选编队 tab); 导入 reports a settings-only import; ←/→ stay inside a focused row's choices.
- screens/cultivation.js (new): CultivationSelects (two compact selects; quiet at the defaults, mint when changed) and
CultivationSection (潜能 1–6 and the four official tiers — effects.json names and effect texts — the rule in one line,
the stand-in note for a chess marked not owned).
- screens/diy.js: an owned 自选 pick's slot card carries its operator's selects; a prototype says it has neither.
- css: the head and the rows share one grid (`--cols`), 32 px quick targets (40 px on a coarse pointer); the detail
slides over the list below 1000 px (was 760 px); below 700 px the row's bonds give way to the name. Headless check at
1920 × 1080, 1366 × 768, 844 × 390 and 667 × 375: no sideways overflow, the head's columns equal every row's.
- The 局内数值 caption says 含潜能与练度 when the numbers carry both (effects.json loaded); the overlay's note says the
skills, modules, 潜能 and 练度 can be set (the operator's in-match level cannot).
- i18n: 21 strings in the four packs (the tier names, short tier labels, the notes, the reset confirm, the import toast).
- Docs: design/match.md (the list, the 潜能与练度 section, 导出's `ops`), ARCHITECTURE.md (screens/cultivation.js).
- Tests: test/ui/loadout-rows.test.js (new, 7): a row's pick / skill slots / modules / aria-pressed / callbacks and the
shared 精锐 preview, its selects (values, options, a change, the changed mark, a stand-in), the head, the detail's
section (tiers from effects.json, the effect text, the stand-in note), a 自选 owned pick vs a prototype, 局内数值 at
潜能1 / 未精英化 and at the defaults with its caption, the css grid and breakpoints. e2e (not run here; their steps
replayed with a headless page): the row pick selector, the phone test (one list, 40 px targets, the detail slides
over, back to the list), the 局内数值 expectations ×1.1, a new 潜能 / 练度 flow (row select → stored, 部署费用 17 / 攻击
454; the detail's 未精英化 → 413; the row follows; 恢复默认).
GitHub PR #281 by @IceCodeNew (this commit is substantially the PR, ported onto 0.2.2-dev; the field-detail range
idea after @BBleae's fork commits c6437f1 / b451bad, credited here instead of the PR's Co-authored-by trailers). The
unofficial skill-range preview the PR first had is gone (the owner's reply of 2026-10-07: only official-like
behaviour); what remains is the range the unit fights with now.
screens/game/range.js inspectRange resolves the open card's unit: an own board piece in prep (its loadout or
stand-in record, its facing — this replaces the prep selection's range effect); in battle or on a scouted board an
ally op / token on the field, at the snapshot's tile, with the live stats' range when there is one; never an enemy,
a dead or undeployed unit, or a scouted bench unit (UnitInfo `area` hand / temp — the tag the server sends since the
#385 fix; renderInfo now keeps it). shared/protocol.js unitStatsEntry gains an optional `dir` (the unit's facing now:
the snapshot tuples carry none and UnitInfo.dir is the facing at send time). tokenVariantFor moves to
gameLogic/loadout.js. Drag, the wheel and the pen clear the overlay.
Added on top of the PR: a test from the server's own scout m.field (board operator ranged, hand / temp none), docs
(PLAYING.md, design/client.md, design/network.md), comments.
Tests: test/ui/inspect-range.test.js (6), test/ui/playtest6_summons.test.js, test/match/unitStats.test.js;
test/ui/inspect-range.e2e.test.js (SP_E2E + CHROME_PATH: run here, 1/1). Golden unchanged (display; m.unitStats only).
- Conflict: docs/PLAYING.md 技能自动释放 bullet — both insertions kept (the 铁钳号 rule from fb7-owner-sim, the #298
pacing sentence).
- Grok review: MERGE AFTER FIXES — the HIGH #298 finding (the cast replaced the due attack) fixed on the branch
(db29acf9: per-skill nextCastAt; golden of #298 alone shrank from 84 scenarios to 2).
- Golden union: 4 matches differ from HEAD though the branch alone left them (coop2-NORMAL-3, coop4-ABYSS-7,
coop2-NORMAL-8-serverrun, solo-HARD-9-boosted): HEAD's side had moved all four (#333's shop RNG …) and on those new
bot paths the merged rules apply — the diffs are in-battle layer gains / damage (卡西米尔 50 → 45, 精准 80 → 76,
叙拉古 176 → 168, kills ±2, solo-HARD-9 LP 397 → 399), no outcome change.
- Broad node --test 5571 (5570 pass, 0 fail, 1 skipped); lint 0 errors; typecheck.
- Grok review: MERGE AFTER FIXES — the end-portal wait (0 s → 3 s, the data's modal crossing) fixed on the branch
(1205336a). solo-NORMAL-1 still fails the Hidden Core (explained in the branch commits).
- Golden union OK; broad node --test 5517 (5516 pass, 0 fail, 1 skipped); lint 0 errors.
- Grok review: MERGE AFTER FIXES — the HIGH finding (skills opened on the device but spending nothing) fixed on the branch
(69d791f9: skills.js allyTargetsOk, per-kit allyTargets flags).
- Golden union OK; broad node --test 5500 (5499 pass, 0 fail, 1 skipped); lint 0 errors.
Substantially PR #301 by @farridge (its non-layout part, ported; the owner's decision of 2026-10-08: 「可以做，具体布局
怎么优化你来决定」 — the card layout is redesigned in the next commit):
- ui/detailPanel.js exports GarrisonBlock; screens/loadout.js LoadoutGarrisons renders the 特质 of the form the detail's
普通 / 精锐 skill toggle shows (garrisons.json `descRaw` with its trigger markers, 「<进入休整期时>」 …) at the top of the
detail; garrisons.json loads with the overlay; its reading sizes in loadout.css.
- ui/loadoutModel.js quickSkillTags: SP recovery (攻回 / 受回 / 自回 / 被动, 部署触发 for a deploy-and-expire passive) and
duration (s, 弹药, 常驻, ∞, 瞬发) from the generated SkillRecord, never from a localized description.
- i18n: the six new strings in the four packs (en from the PR; ja / ko / zh-TW added).
- Tests: test/ui/loadout-quick.test.js (from the PR, 4): quick tags of 空弦 / deployment skills / ammo / infinite, the 特质
blocks of 空弦 / 星熊 / 缄默德克萨斯 both forms; loadout-stats.test.js reads the new import.
Seen: 蛇屠箱 and the seven other operators with the range 0-1 (角峰, 古米, 泡泡, 折桠, 菲莱, 塞雷娅, 余 — 16 chess records, all
melee blockers with block 3, the elite 角峰 4) block and hit only on the tile they stand on. The layout planner maximizes an exposure model whose
kill estimate saturates in a full layout; the tile then falls to a 1e-6 rng tie-break, so a guard stood off the road, where
it neither blocks nor hits (reproduced: planLayout with six of them on act2autochess_m04, seed 11 — 蛇屠箱 off the road; over
40 seeds of the eight stages 16 of 320 guards stood off the road: 11 because the eight guards outnumber the 7 free road tiles of
act1autochess_m03 / act2autochess_m02, 5 avoidably — after the change only the 11 are left).
Official: excel range_table "0-1" is the single grid (0,0).
Now: planLayoutSteps keeps the best road tile (model.ground) for a blocker whose deployed range grid (rangeRec: its
loadout's) is exactly one (0,0) grid, and takes it before the overall best; with no free road tile the best legal tile as
before (a high-ground 「可以放置于远程位」 blocker keeps its own preference first).
Not taken from the PR: the scarce-card penalty of buyScore, the item scoring (人事部文档 "only with nine 炎" contradicts the
official 部署上限变为 9), the 神秘 morph fit and the draft yield — AI heuristics with no official text, and the golden
swing they caused (a solo match that cleared the Hidden Core no longer did) came from them, not from the placement.
Golden (matches only; every other family unchanged, --twice identical): solo-FUNNY-1 (from R7: boss pool left 74740 -> 71550,
still a loss in R8), solo-NORMAL-1 (from R4: same win, Hidden Core cleared, team LP 26), solo-HARD-1 (from R4: boss pool left
3046117 -> 2940542), solo-ABYSS-1 (from R4: 2303334 -> 2281356, still a loss in R13), coop2-NORMAL-3 (from R4), coop2-HARD-4
(from R4), coop4-FUNNY-6 (from R3: virtualMs 128500 -> 127000), coop2-NORMAL-8-serverrun (from R3), coop2-NORMAL-70-diy (from
R6): rows, stats, lineups and rng draws differ, wins, losses and rounds passed do not.
botbench A/B on 12 seeds each (solo HARD / ABYSS, coop4 NORMAL / HARD): wins 1 / 0 / 11 / 3 both before and after, rounds
passed 13.00 / 11.92 / 14.00 / 13.00 -> 12.92 / 11.92 / 14.00 / 13.17, LP left 18.7 / 6.5 / 80.1 / 27.9 -> 18.8 / 7.4 / 83.7 / 28.0.
Tests: bot.test.js — the PR's saturated-layout case (蛇屠箱 blocks and hits only on a road tile, a real battle on both tiles;
the road full: a legal fallback) and every stage over 40 seeds (the eight guards fill min(8, free road tiles) road tiles);
both fail on the parent.
- Conflicts: the stats entry (merged fb7-stats) and the settings button both sit in the lobby / room top bars — both kept;
public/js/ui/settings.js (#321 diagnostics, voiceLang, SettingsButton); docs/PLAYING.md table rows resolved line by
line against the merge base (#368, #354, #333 edits kept); docs/SIM.md / design/network.md snapshot field lists
(resolved + ammo / wolves / neg / stand / standCut); design/client.md Screens line (#368's AI switch kept).
- Grok review: MERGE AFTER FIXES — the shared-field capsule finding fixed on the branch (1a880699).
- Golden: roster 23, bonds 2, fields 11, matches 18, diy 11 — #308 totals / snapshots and #333's shop RNG (union OK);
broad node --test 5473 (5472 pass, 0 fail, 1 skipped); lint 0 errors; typecheck; i18n strict clean.
Grok's review of the 铁钳号 change (HIGH): the device opened every non-heal skill, but kits that pick their victims
with enemiesInKeys only were spent on nothing — 蕾缪安 S3 went active with 0 locks, 艾丽妮 S3 and 陈 S2 struck nothing,
引星棘刺 S2 threw at the empty far tile. Audit: every skill of every chess (loadouts) and of the 自选 pool, placed with
only the device in range; of the 351 skills the device opened, 67 never touched it.
- skills.js `allyTargetsOk`: the device counts for a skill's start only if the skill acts on it — SkillSpec
`allyTargets` (true / false) from the kit, else [ASSUMED] a skill acting through the unit's own attacks (timed and
still attacking, or a cast that is the next attack) on a unit that attacks. An onStart-only cast or a no-attack timed
skill picks enemies itself and waits for one (德克萨斯 S2 剑雨, 锡人 S1/S2, 铃兰 S3 … about 50 skills).
- Spend on the device (allyTargets: true; its kit cancels the hits, its 受击回复 counts): 蕾缪安 S3 locks it and spends
her bullets (ammoUsed, so the per-bullet layers come) and shells it; 艾丽妮 S3 strikes it; 陈 S2 draws on it after
every enemy; 引星棘刺 S2 throws onto it and its pulses hit it. queries.js allyTargetsInRadius for their areas.
- Explicitly closed ([ASSUMED], allyTargets: false): 普罗旺斯 S2 (never shoots above 80 % HP), 卡涅利安 S1 (charged: no
attack), 山 S2 (blocked enemies), 余 S2 / S3, 迷迭香 S3 (blocked only), 荒芜拉普兰德 S1 / S2 (drone locks), 淬羽赫默 S1–S3
and 遥 S1–S3 (heal attacks), 机械师 S3, 珊比 S2 / S3, 陈 S3 / 艾丽妮 S2 (their guards already refused).
- 雪猎's empty-magazine cast back to enemies only (its S2 does not act on the device).
Re-audit: 281 skills open on the device alone and all act on it (蕾缪安 S3 through her locks); none of those that
acted on it before stopped. Docs: SIM.md, design/engine.md, PLAYING.md.
Tests: test/content/feedback7-allytarget-kits.test.js (5; all fail on the parent).
The owner's decision of 2026-10-08: 「调配干员里自己设置吧，默认满潜满加成」. Official: the account's potential (调度手册);
the 自持有 bonus to owned operators of the same name, initial and elite alike (activity_table cultivateEffectList).
- shared/protocol.js: room.loadout { entries, ops? } — `ops` = { [charId]: { potential?: 1–6, cultivate?: 0–3 } }
(isLoadoutOps, ≤ 256), checkLoadoutOps (strict like checkLoadout: an operator of the 干员调配 roster or the 自选 owned
pool — cultivationCharIds; defaults dropped, entries stored complete); absent = none set.
- server/lobby.js / net.js: checked with the entries (either refused ⇒ nothing stored), session.ops / seat.ops /
seats[].ops, match.setLoadout(playerId, loadout, ops) in INFO_CHECK only.
- PlayerState: ops re-checked and frozen (bots none); cultivationFor (shared/potential.js cultivationOf: null for a
补位 stand-in or a prototype 自选 pick); loadoutFor carries potential / cultivate; battleInput states `potential` /
`cultivate` for every owned operator (the chess as itself, an owned 自选 pick) — an old seat without ops and every bot
at 6 / 3; the 自选 summons' hand count follows the owner's potential (望's 棋子 6 below 潜能3); m.private.ops; the
scouting UnitInfo carries potential (below 6) / cultivate.
- tools/golden.mjs: the direct-spec scenarios field the default player (owned operators at 潜能 6 / 练度 3; stand-ins
and prototype picks neither), as PlayerState does.
- Golden: every battle with an owned operator moves — ATK / DEF / max HP ×1.1 from the first second (roster-001:
角峰 ends at 3527 HP instead of 3115; the first differing snapshot window is 0–10 s in every moved battle); the
matches part at round 1–8 (battle results, then the bots' boards). Unchanged: the 10 stand-in battles (a stand-in has
no 练度) and diy-135…138 (the four prototype 自选 battles). No engine error.
- Docs: design/match.md §16 (潜能 / 练度), design/network.md (room.loadout ops).
- Tests: test/match/potential.test.js (new, 4): the ops wire / data check, battle input of a human (settings, defaults,
stand-in none, elite shares the charId), an old seat and a bot at the defaults, spec → battle, scouting units; 自选
owned pick vs prototype and 望's hand count; Match.setLoadout in INFO_CHECK only, a refusal changes nothing.
test/lobby-loadout.test.js (+2): strict validation, storage beside the loadout, an older client's message, the real
match's briefing (m.private.ops) and lock. test/match/runner.test.js / diy-shop.test.js expect the new fields.
Seen: the Art 教鞭 (杜宾's strategy) drew three 战术特训 cards and added a random one — a documented simplification (docs/META.md
§2.5: a personal choice would need its own phase, protocol message, timer and AI takeover). Official: act2autochess
choice_event hunter_band_1 is a PERSONAL_CHOOSE; PRTS 下半 记录 §法术 教鞭 "使用后销毁，于3个战术特训的悬赏任务中选择一项". The
owner decided on 2026-10-08 to follow it. Taken from PR #353 (code and tests by @Sukvii), ported to the split 0.2.2 modules.
Now: ctx.offerBountyChoice -> Match.offerBountyChoice keeps up to three different mode-eligible cards (15 of 20 in 标准 /
险境, 20 in 绝境 / 终极; one meta-rng shuffle) in PlayerState.personalChoice, m.private only (same cards after a
reconnect); g.choice { idx, choiceId } confirms it (Match.pickPersonalChoice: owner, id, round, index; a request without
a choiceId is still the public draft, a stale id never falls through to it). No new phase or timer: Ready is refused while
it is open, the prep's deadline picks at random before the temp pieces and Ready, a bot or a 托管 seat picks with the bounty
scorer (botPickCard, the bots' seeded rng; at the prep's start, after the cast, before Ready; the scheduler's fallback
picks at random if a step fails). The client reuses ChoiceOverlay (personal: no order, no turns, the prep's countdown) over
a locked prep, the Ready reasons, the mock harness; six UI strings in en / ja / ko / zh-TW. An eliminated player's offer
goes with it; invariants.js flags an offer outside its owner's own open prep.
Not taken from the PR: its §25.24 in the released docs/history/0.2.0.md (the 0.2.2 section is the integrator's), its edits
of the Ready toggle beyond the personal-choice case, its tail placement of the i18n keys.
Golden: coop2-NORMAL-8-serverrun (ai_0 杜宾 casts 教鞭 from R2 on and now scores the three cards: teamLp 39 -> 34, virtualMs
121400 -> 118400, rng shop 511 -> 483 / bots 18063 -> 18070 / meta 112 -> 103) and coop2-ABYSS-10-boosted (ai_1 杜宾: virtualMs
121650 -> 124650, shop 590 -> 589 / bots 22075 -> 22096 / meta 141 -> 154); both first differ in round 2, every other scenario
and family is unchanged; node tools/golden.mjs --twice is identical.
Tests: items.test.js (three distinct cards, the tapped one applied once, nothing before the confirm, a refused cast keeps
the Art / count / rng / uid, an eliminated player's offer), economy.test.js (Ready refused, the deadline picks before
endPrep, solo prep untimed), connection.test.js (two players' offers, reconnect, foreign / duplicate / old ids, an offline
player's pick at the deadline), bot.test.js (托管 takeover incl. the failing-step fallback), bot-smarter.test.js (the scorer's
pick, deterministic per seed, no rng draw when nothing is pending), followups.test.js (R14 / R15), draft.test.js,
unitStats.test.js (protocol), hardening.test.js (invariant), gameLogic / playtest3 / playtest4 (view model, Ready reason,
ChoiceView, actions), mock.e2e.test.js and playtest2.real.e2e.test.js (browser, SP_E2E=1).
The owner's decision (relayed 2026-10-08): only the 速射手 air priority outranks the attacker's own blocked enemy — PRTS
索敌的概念 「阻挡（近战限定）→特殊优先级→…」 — while the user's playtest #6 rule 「阻挡了就一定要能打到」 keeps 'blocked first'
for every other attacker and priority. PR #220 by TsangAsuna (for #205) moved every special priority ahead of the block;
re-implemented for priority 'fly' only (targeting.js sortEnemyTargets: the fly key, then blocked, then the rest as
before) — the trait's priority and a skill's targeting priority 'fly' (新约能天使) alike. Seen (repro): 能天使 on a melee
tile blocking a walker kept shooting it with a flyer in range; now the flyer, and 远牙 (神射手, lowest DEF first) still hits
the enemy it blocks over a softer or a flying one. Docs: SIM.md (blocking), design/engine.md, PLAYING.md 阻挡.
Golden: roster-026 (流星, 速射手) and roster-037 (新约能天使's priority 'fly') now shoot flyers in range while blocking.
GitHub #375 (FrogThai): 「水图boss会从下面这条路往蓝门走，实际应该从上路走」. The walking leaders of 战场#08(下半) 涨潮控制
(act2autochess_m04) are 铳 (level_act1autochess_h07_02 route 6; the Hidden Core 铳 on act1autochess_h08_02's patrol
routes) and 卢西恩 (act2autochess_h07_05 route 8): PATROL_MOVE (3,9) → (2,3) → (5,9). Measured (notes repro-375.mjs):
ours 「3,9 2,9 2,6 4,6 4,4 2,4 2,3」 — the lower road through the 深水区 (3,6) / (4,6) — against the official SPFA
「3,9 5,9 5,4 2,4 2,3」 along row 5 with one 深水区 (5,6); the mirrored (3,11) → (2,17) split the same way.
Cause: grid.js _buildField takes the blockable-ground preference pointer (user playtest #2) when its route crosses
strictly fewer non-blockable tiles; 深水区 is non-blockable itself (no deployment), so the lower road counted 2 against
the upper road's 3 (two floor tiles and one water tile) and won. Official pathfinding is the plain 4-neighbour SPFA
(research 08 §3.1). Now the preference pointer is taken only when its route also crosses no more 深水区 than the official
pointer's (a per-tile `deep` count beside `cost`, start excluded). Rule from the Grok triage (scratch-i1b/375.patch),
re-implemented with a precomputed Grid.deepWater mask (terrain 'deepsea': tile_deepsea and tile_deepwater).
Scope (notes dump-fields.mjs / check-official.mjs, every stage × normal / 联防 / boss rect × every walkable
destination): 48 fields change, only on the two water maps (act2 m04 boss field; the inactive act1 m05), 149 pointers,
every one back to the official pointer. The 21 accepted DEVIATIONS (战场#01 / #02 floor, #07, the inactive #05's 7
water-avoiding routes) are unchanged, as are all 154 audited gate routes and data/stages.json's groundPaths
(tools/build-data.mjs --offline into a scratch dir: every data file identical). Other boss-field preference routes of
m04 that cross fewer 深水区 than the official one stay.
Tests: test/sim/pathing-official.test.js — two new #375 tests (the four patrol legs equal the pure official waypoints;
in a real battle the 铳 climbs to row 5 and steps on the 深水区 (5,6) only), both failing on the parent;
test/sim/pathing-crosscheck.test.js — the independent reference port gets the same 深水区 rule (it flagged both the
patrol leg and a random-crate normal-field route of act2 m04 until then); test/sim/pathing-blockable.test.js — the WALK-leg
audit compares with the equal-length routes that cross no more 深水区 than ours (the 14 accepted arena exceptions
unchanged). Grid-related suites 777/777.
Golden:
- fields boss-boss_2-pair (铳 pair on act2autochess_m04, act1autochess_h07_02): the two 铳 patrol the upper roads;
leader left 5311560 → 5301010, p1 damage up / p2 down, deaths 0 / 2 → 1 / 3; still 8 / 8 kills.
- matches coop2-ABYSS-10-boosted (act2 m04; Final Assault 卢西恩, Hidden Core 铳): still cleared, team LP 800 → 615. The
bots' operators stand on the lower rows (the same tiles as before); with the leaders on the upper road the Final
Assault lasts 335 game s instead of 247 and the Hidden Core 638 instead of 190, both past the 150 real-second mark:
the overtime drain takes 17 + 168 LP (none before), no leak.
Applied the host-only room option 「AI 队友最后选择」, default off. Off, the draft order and the random streams stay as they are. On, humans draft before AI seats after the existing shuffle, and a skip stays ahead of the AI seats.
Dropped docs/history/0.2.2.md and every §27 citation. The player-facing sentence is in docs/PLAYING.md. The room-option rule is in docs/design/match.md, docs/design/client.md and docs/design/network.md. The #338 per-bot faction and ban presets are not included.
Tests: node --test test/lobby-ai-last.test.js test/match/ai-picks-last.test.js test/ui/ai-picks-last.test.js — 18 pass. node --test test/lobby.test.js test/client-static.test.js — 378 pass. node tools/i18n.mjs check --all --strict — 1032/1032 in en, ja, ko and zh-TW, 0 errors.
Community report relayed by the owner: 圣约送葬人 hitting 白铁's device spent no bullet and gained no layer.
Our attacks already selected the device (Battle.setAllyTarget, damage cancelled), but every skill start
condition counted enemies only, so her MANUAL ammo skill (DEFAULT) never opened. The owner's rule of
2026-10-08: a registered ally target in range counts like an enemy for a non-heal skill's auto start.
- skills.js `_allyTargetIn`: DEFAULT (about to attack, the initial range; SEARCH / ACTIVE_RANGE through
it), CUSTOM_RANGE, SKILL_RANGE, GDGLOW_SKILL_2 (anywhere) and the content trigger ranges. Heal skills
never count it (禁疗). It stays out of enemiesInKeys / canTargetEnemy.
- 雪猎's empty-magazine S2 cast counts it too; allyTargetsInKeys takes a Set of tile keys.
- 白铁™多功能平台 stays no target. Kits whose own selectors pick enemies (陈 S3, 艾丽妮 S2 guards,
蕾缪安 S3 locks) are unchanged.
Tests: test/sim/feedback7-allytarget-cast.test.js (4; 3 fail on the parent).
Seen (reproduced on the parent, notes repro-354.mjs): band_orchid's active refresh copies the top-tier operator into another
slot and freezes the copy (bands/meta.js band_shop_refresh_copy_max_lv_char, slot.frozen = true) while shop.frozen — the
toggle — stays false; m.private sends both (views.js: shop.frozen and slots[i].frozen). The shop bar passed the toggle
(`frozen = !!shop.frozen`) to every ChessCard / ItemCard, so no card showed the frost: the server kept the card to the next
prep, the player could not see which one.
Now: shopBar.js slotFrozen(slot, toggle) = the slot's own flag, or the toggle (a frame without per-slot flags — the dev
mock — and the old behaviour for the 冻结 button, which still copies the toggle onto every unsold slot); both card kinds
use it. The bar's frame (is-frozen) and the 冻结 / 解冻 button stay on the toggle. The rule is not touched (bands.test.js,
and the new test): a second manual refresh re-rolls the frozen card too — like the toggle, "再手动刷新会重抽" — and
freezes one of the new pair; with no further refresh the frozen card stays in the next prep (the flags are cleared
then, round.js startRound).
- Tests: test/ui/feedback7-orchid-freeze.test.js (4: the wire — exactly one frozen card of the pair, toggle off — and the
cards drawn from it; the toggle still freezes every slot; the rule unchanged across two refreshes and a round; the
ShopBar source pin); fails on the parent (no slotFrozen, the cards passed `frozen`).
- Docs: PLAYING.md (冻结 row), design/network.md (m.private slot.frozen). Golden: unchanged (client render only).
Applied unchanged after review (Grok static review: MERGE; the diff read in full: no network, no dependency).
UnitView._updateImpostor refreshed a unit when (frame + phase) % interval === 0 with a random phase per unit, so the
refreshes of a frame were binomial: reproduced on the parent with 7 pen figures at interval 3 — 5 1 1 5 1 1 … and
2 4 1 2 4 1 … refreshes per frame instead of 2 / 3. Now render/app.js hands every impostor unit the next slot of the frame
(ctx.impostorSlot, a counter reset in frameBody), the unit refreshes when (frame + slot) % interval === 0, and a unit
whose slot keeps moving with the frame (the units before it culled on and off in step) is still refreshed after 2
intervals (imp.last); a context without slots keeps the random phase. Each unit's rate and the total work are unchanged;
no sim or pixel-as-state change.
Step 2 of #327 (a time budget by priority) is not part of this and still needs a decision.
- Tests: test/render/unitview.test.js 'impostor refresh slots' (3: slots in update order — fails on the parent, the
starvation cap, the no-slots context); unit view / render suites unchanged.
- Docs: design/client.md. Golden: unchanged (client only).
Seen (Grok's review, reproduced here with the real sim): the sim's deploy emits a deployment-activated skill's
['skill', id, 1] before the unit's ['spawn', unitInfo] (宴: three events earlier) and audio.js dropped a skill event of an
unknown unit, so that one cast made no sound while every later one did.
- audio.js: pendingSkill — a 'skill' ON for an unknown unit is held (one per unit, 64 at most, the oldest go first);
_track plays it once when the unit's spawn arrives; a skill OFF is no cast and is not held; setFieldUnits drops the holds.
Changed from the PR: setFieldUnits clears the holds BEFORE it tracks the new field's units (the PR cleared after, so a hold
left over from the previous field would have sounded for a unit of the new field that shares its id — ids are per battle).
- screens/game.js: NOT the PR's change. The pre-entry buffer (keepEarly: every spawn, die, deploy, status and skill since the
field's m.field, up to 1500 of them) is handed to the sound as its 'spawn' tuples only (new early.js audioEarly): the whole
list would play every old death, deploy line and cast at once on entering. The live path (onEv) already forwards the full
list, so the hold alone fixes the live silent first cast; a deploy-tick cast that falls into the short pre-entry gap
stays unheard (a deliberate gap — replaying "only the casts before their spawn" would still sound a minute-old cast when a
field was buffered while another one was on screen).
- Tests: test/ui/audio.test.js (3: the held cast in one batch and across batches, once, no hold for a skill OFF; a hold does
not outlive its field and is bounded; 宴's real deploy: the sim's cast comes before the spawn and sounds exactly once — all
fail on the parent) and test/match/feedback1-runner-forms.test.js (the audio gets the spawn tuples only: silent replay that
still teaches the units, the next live cast sounds; the whole list would have sounded old cues). Docs: design/client.md.
- Golden: unchanged (client audio only).
Applied unchanged after review (Grok static review: MERGE; the diff read in full: no network, no new dependency, no
shipped eval). The leak was real on feedback7: mountFieldView raced createFieldView against a timeout and, on failure,
removed every child of the host while the factory kept running — Pixi finished later and appended its canvas over the
simplified view (taking its clicks); the host wipe could also delete a newer view mounted by the same hook.
- ui/fieldHost.js: mountFieldView(host, { signal }) owns an AbortController; the timeout, a failed import or an invalid
view aborts it, a view a factory returns after that is destroyed, an outer abort (useFieldView's cleanup) rejects
without a fallback; nothing clears the host's children any more.
- render/app.js createFieldView takes the signal: the Pixi load / manifest / font waits end on abort, the synchronous
build registers its disposables (setupCleanup) so a throw releases the app, tiles, fx, impostors, listeners and
observers it made, the board-art and 3D-board waits moved after the view exists (aborting them is view.destroy(),
idempotent), destroy() also removes the canvas. render/app/host.js withTimeout(p, ms, signal).
- Tests: test/ui/fieldHost-lifecycle.test.js (5, node vm, fake timers: all fail on the parent) and the opt-in real-browser
test/render/initialization.browser.test.js (6 scenarios in headless Chrome, 5 fail on the parent: a late Pixi load
covering the fallback — with a factory that honours the signal and one that ignores it —, art timeout, a setup
exception, an unmount racing a newer mount); test/ui/playtest2.test.js's source pin takes the new indentation.
fieldview / browser suites in Chrome unchanged (the two 4× throttled fps tests flake when two Chrome files run
at once, pass alone).
- The createFieldView body stays unindented inside its new try (the PR's choice, to keep the diff reviewable).
- Docs: design/client.md §9. Golden: unchanged (client lifecycle only).
The three PRs patched the same snapshot and HUD lines (Grok's review: do not merge three diffs), so this is one design
built from their code and tests; IceCodeNew is the commit author (#303, #378), Convey123 co-author (#383).
- b.snap gains three optional, display-only lists beside the unchanged 9-field tuples (snapshot.js ammoView / wolfView /
negView, called from battle/events.js snapshot(); a living, deployed, visible ally, only when it has one):
ammo [[id, rounds left, magazine]] (whole rounds; the magazine is the activation's real total that unitTuple's SP
fraction runs on, extras and refills included), wolves [[id, 狼影 left, talent maximum]] (mem.shadows of the new
mem.wolfCapacity, set by tokens.js wolfPack and the managed vigil kit), neg [[id, fill]] (unit.negFill(), a function
op-hsgma2.js sets: the pool is read when the snapshot is taken, no per-tick sync). UnitInfo.ammoSkill (an ally whose
skill is an ammo magazine) is known from the first appearance: fx events run before the frame's sample, so the aura
cannot wait for the first ammo row.
- render/interp.js keeps the lists per snapshot (malformed rows / unknown ids dropped) and sample() hands them out from the
older snapshot like flags (they step with the skill flag); render/units.js draws them: the ammo magazine replaces the
SP bar (a cell per round, thin cuts, exact fill, no cuts under 2 px a cell), a row of diamonds under the bars for the
wolves (the element gauge row moves below), the red bar over the drained HP bar for the pool. An ammo skill draws no
standing aura (rings.js _aura; the activation burst stays).
- PR #378's wolf-head avatars are NOT taken: the count is data (the talent text 至多3只, mem.shadows), the HUD art of the
official counter is in no game data we have, so the pips are plain diamonds [ASSUMED look]. The PRs' own tests are
folded in and extended (test/sim/hud-readouts.test.js, test/render/hud-readouts.test.js, op_hsgma2.test.js).
- Docs: network.md, SIM.md §9, client.md, PLAYING.md, types/core.js.
- Golden: unchanged (283 scenarios; the canon hashes units / down / elem only, no sim state moved).
- Tests: 4 + 14 new + 1 in op_hsgma2 (12), all fail on the parent; 344 related tests pass; lint 0 errors, typecheck,
check:imports, docs tests.
Co-authored-by: Convey123 <Convey123@users.noreply.github.com>
sganggs