Conflict: server/match/player/basics.js imports (W5G's checkNotOwned + W5I's msg / dn). data/i18n/en.json rebuilt on the merged data (tools/build-i18n.mjs --offline): the 自选 data adds texts of CN-only operators and modules the EN client lacks, so coverage is 97.6 % (skills 95.9, talents 95.1); the coverage floors in test/i18n-data.test.js and docs/I18N.md say so. Golden unchanged; i18n / lobby / match / ui suites 1144/1144 after the rebuild, i18n e2e 1/1.
Conflicts: tools/golden.mjs / test/golden/README.md take both new families (standins from W5G, diy from W5H); DATA.md / SIM.md keep both texts. data/ equals an offline build of the merged tools/build-data.mjs. Golden regenerated on the merged code: against feedback5 nothing moved, diy.json is new; against fb5-diy the W5G / W5J / W5L scenarios moved, and diy-004 / diy-006 (the prototype battles with Touch, changed by W5L) — diy-001/002/003/005 identical.
public/i18n/en.json: Chinese msgid -> English, seeded from PR #70 (@YuriRestia, credited) for every Chinese
literal of the current code (≈830 entries; 966 of the 1 036 literals left for phase 2 already translated),
completed by hand in the official EN terms. public/js/ui/lang.js: boot language (?lang=en|zh, then
localStorage sp.pref.lang, else Chinese; the browser language is not consulted), switchLang, useLang,
LangToggle; main.js renders after initLang and re-renders in place on a switch, and translates server
texts (m.toast / m.ticker frames, room.closed reasons, error codes via ui/toasts.js describeError).
Converted: title screen (English title, wordmark hidden), lobby, room, settings (语言 row, On/Off,
quality, hotkeys), the shell (crash screen, toasts) and DifficultyTag.
tools/i18n.mjs: extract (msgids + untranslated literals per file), check (msgids missing from en.json),
codemod (html`` text / attributes / strings / template literals -> t(), nested strings, N_ for module
level, i18n-ignore lines), seed (fill en.json from a msgid map).
shared/i18n.js: gettext-style t() / tc() with the Chinese text as msgid (untranslated -> Chinese), {name},
{0} and {n|one|other} params, lists, game-data names (dn / tName), setLang / getLang / onLangChange, and the
server message wire format (msg -> wireMessage -> translateWire). Chinese stays the default, English is a
switch (the owner's decision of 2026-10-05).
tools/build-i18n.mjs (build-data.mjs untouched) walks the zh_CN and EN tables side by side and writes
data/i18n/en.json: the official English of every data/*.json text by record id and field. Blackboard texts
are matched by template (the remake's numbers read back and filled into the English template); stage names
are composed; what the EN client lacks falls back to tools/i18n/fallback-remake.json and fallback-pr70.json
(PR #70 by @YuriRestia). Default EN source ArknightsAssets/ArknightsGamedata (has act2autochess);
Kengxxiao/ArknightsGameData_YoStar is archived (2025-11) and has no 盟约 season (--source yostar).
Coverage: 14 922 / 14 931 texts (99.9 %), 99.3 % official; only 9 CN-only 甄选 operator names stay Chinese.
shared/i18nData.js: the overlay format (index-aligned leaves + a 2-char hash of each Chinese source, so a
stale leaf is never applied). public/js/data.js: setLocale('en') downloads the overlay once; get / lookup /
list return localized copies; getRaw keeps the Chinese file; localeName maps names.
Downloaded 1,295 files, 127.2 MB (133,426,089 bytes) into the shared public/assets; no existing file changed or removed. data/assets.json only gains entries: 72 chars (all with avatar, portrait, Front and Back Spine), 38 tokens (avatars; no Spine — upstream has skin variants only), 208 skill icons, 10 sub-profession icons, 102 SFX units, 131 module type icons.
A new family (test/golden/diy.json): no existing scenario moved (roster 49, bonds 46, fields 22, matches 16 identical); determinism checked with --twice.
Golden regenerated on the merged code (the textual merge had no conflict and equals the regenerated files): against feedback5 only the new match coop2-NORMAL-14-standins and the new standins.json (10 battles); against fb5-standin-game exactly the W5F and W5J scenarios (roster-001/002/007/009/016/026/027/028/038/040/041/042/044, bond-egirShip-low / siracusaShip-high / suntShip-high, boss-boss_2-pair / boss_6-pair, solo-NORMAL-1, coop2-NORMAL-3, coop3-ABYSS-5, coop4-FUNNY-6, coop2-NORMAL-8-serverrun); standins unchanged. test/match 536/536, the stand-in / golden / docs suites 175/175.
backups.json 171,616 → 1,884,868 bytes (gzip ≈ 142 KB); every other data file is byte-identical (buildTokens' record head moved into summonRecord, shared with the DIY summons). No golden scenario moved (none fields a 自选 piece).
Golden regenerated on the merged code (only test/golden/matches.json conflicted): against feedback5 it moves exactly W5F's scenarios (roster-001/002/016/026/027/028/038/041/042/044, bond-egirShip-low, bond-suntShip-high, boss-boss_2-pair, boss-boss_6-pair, solo-NORMAL-1, coop2-NORMAL-3, coop4-FUNNY-6, coop2-NORMAL-8-serverrun); against fb5-bugs3 exactly W5E's (unite-1/2 and the 8 match scenarios of the 联防 map and the full-hand rule). data/ equals an offline build of the merged tools/build-data.mjs.
tools/build-data.mjs resolveTrigger: a MANUAL operator skill on DEFAULT (no row, or a DEFAULT entry of the deviation
tables) or on the SEARCH row ("在初始攻击范围内存在敌人时释放技能": 解放者 / 阵法术师 / 安洁莉娜 — ACTIVE_RANGE_OVER) whose
running range strictly contains its own is ACTIVE_RANGE on that range, rawRule the official row; validateAll accepts a
DEFAULT deviation resolved to ACTIVE_RANGE (chess and stand-in tables) and ACTIVE_RANGE over SEARCH. Data rebuilt
offline, JSON-compared: only skills[].trigger rule / customRangeGrid of 薄绿 S1 (x-2), 蜜蜡 S1 (x-2), 卡涅利安 S3 (x-2),
玛恩纳 S2 (2-3), 安洁莉娜 S3 (y-4) — rawRule SEARCH — and 深巡 S2 (3-2, rawRule TAKE_DAMAGE), normal + elite (32
leaves); backups.json unchanged. Kits: 薄绿 S1 drops its DEFAULT trigger and 深巡 S2 PR #12's DEFAULT line, both read the
data now (雷蛇 S2 keeps PR #12's line). The sim needs no change (SEARCH was the DEFAULT condition on the initial range
every tick; ACTIVE_RANGE is the same on the running range). Docs: DESIGN §5.6 and the §21.20 flip row, DATA, SIM,
PLAYING, kits README, shared tier comments, skills.js header.
Tests: data (78 records / 39 skills, the five from SEARCH, the SEARCH skills that stay), feedback5-active-range (安洁莉娜
S3 on its y-4, S2 keeps SEARCH), feedback1-tank-triggers and kits_t1t2 (深巡 S2 on its 3-2, before any hit), kits_alt_t3
(薄绿 S1 on its x-2), feedback1f-fence (the idle measure uses the interval of the cooldown still running: the S2 now casts
between two attacks), docs-consistency (one PR #12 kit line left).
Golden: roster-001 / 002 (深巡), roster-016 / 044 (玛恩纳, 安洁莉娜), roster-026 / 027 (薄绿), roster-041 / 042 (卡涅利安),
bond-egirShip-low / bond-suntShip-high (深巡), boss-boss_2-pair / boss-boss_6-pair (玛恩纳, 安洁莉娜), matches solo-NORMAL-1,
coop2-NORMAL-3, coop4-FUNNY-6, coop2-NORMAL-8-serverrun — each fields one of the six skills, which now casts when an
enemy enters its running range.
- simdata: DataSource keeps `backups` (rawBackups follows the fallback chain); getStandIn = shared/standIn.js
standInRecord normalised (the chess's ids / bonds / 特质 / tier, the stand-in's body, def.standInFor); a PRESET / DIY
chess or a source without backups fields itself; the per-battle view maps id-only lookups.
- spec.js keeps `standIn` only when exactly true (never on a token); Battle passes it to getChess; syncUnitLoadout
carries it; UnitInfo `standInFor`.
- content/index.js kitOf: a stand-in's kit is KITS[def.charId] only; no default skill (a stand-in kit writes `skills`);
without a kit the generic kit plus genericTalents. kits/index.js STANDIN_KIT_FILES / STANDIN_KITS, merged after the
tier groups (old keys keep their order; golden unchanged).
- server/data.js and tools/setup.mjs expect backups.json.
- test/content/standin.test.js: the plumbing and the eight 预备干员 on all 28 chess records they replace.
The owner's decision of 2026-10-05 (the approved 补位 plan). Keyed by the stand-in's charId, because a form belongs to
no chess; validateAll fails the build when an entry stops meeting a TAKE_DAMAGE row. build-data --offline changed only
those 15 trigger rules in data/backups.json.
A deliberate deviation from the official 技能策略 (community report: skills with a larger range waited for the normal
range). tools/build-data.mjs resolveTrigger: a MANUAL operator skill still on the basic strategy (no class row, no
TRIGGER_DEVIATIONS entry) whose running range (its rangeId grid, or the own range grown by
ability_range_forward_extend; never 被动效果：攻击范围扩大) strictly contains the operator's range gets rule ACTIVE_RANGE,
customRangeGrid = that range, rawRule the official DEFAULT; validateAll checks it. server/sim/skills.js: ACTIVE_RANGE
checks the DEFAULT condition (a targetable enemy or a blocked one; a heal skill an injured ally) on that range, grown by
the permanent rangeExtend, every tick. Data rebuilt offline: only skills[].trigger changed (33 skills, 66 chess records,
9 stand-in form records). 深巡 S2 (§21.29), 重装 TAKE_DAMAGE, the SEARCH / CUSTOM_RANGE / SP_FULL rows, 烛煌 S2 / S3
(not containing her 3-1) and 引星棘刺 S3 keep their rule. Docs: DESIGN §5.6, DATA, SIM, PLAYING, kits README, notes.
Tests: new test/sim/feedback5-active-range.test.js; data, playtest6-skills (银灰 S3), kits_alt_t2 (蒂比 S1), kits_t5
(百炼嘉维尔 S2) updated; feedback1-transform's fixtures move 莫斯提马 one tile (her S3 now casts on its 3-15).
Golden: 27 roster, 25 bond, 21 field and 14 match scenarios moved — every one fields an operator with one of the 33
skills, which now casts when an enemy enters its running range instead of the initial range.
tools/package.mjs packs git-tracked runtime files and production dependencies. --lite leaves out local art, fonts and the local asset manifest. --dry-run lists the files and the uncompressed sizes and does not write a zip.
tools/check-imports.mjs lists boundary hits and exits 1 only with --strict. The current hit is the Node data loader's node:fs, node:path and node:url. server/data.js stays allowed.
Flat config for Node and the browser, with unused names as warnings. The run is clean of errors after three small fixes: the lobby spectate path imports ERR, a disable comment for a rule that is not installed is gone, and two atlas regexes spell the BOM as an escape.
chess.json: every record carries the shop row's backupCharId / backupTmplId / backupCharSkillIndex /
backupCharUniEquipId / backupCharPotRank verbatim as `backup` (both forms; nothing else changed).
data/backups.json: units = the 17 原型干员 with a form per status they fight at (43 forms: stats, trait,
talents, every unlocked skill, modules), diy = the 4 slots, the prototype picks per tier (9 6★ at 5 and 6,
six 4★ at 5), the 78 owned-6★ picks and their faction bonds (mainPower + subPower vs powerIdList, else 协防).
shared/standIn.js composes a stand-in (or a DIY prototype) into a chess-shaped record; the build fails when
the form builder cannot give back every PRESET chess field for field. test/backups.test.js (10 tests).
The 联防 phase plays m_bat_corrosion (intro + loop), the track of the official escaped stages, instead of carrying on with the round's battle music; without the downloaded track it falls back to the battle music. Asset plan, manifest entry (audio.bgm.unite), audio.js and tests. Contributed by @Convey123.
The trait line is on every hookmaster and is not trusted. Placement is the loadout at place time: golden 歌蕾蒂娅 carrying uniequip_003_glady may use a ranged tile; everyone else stays on the ground.
The owner's decision (2026-10-03, after community feedback): 深巡 S2 行动能力剥夺, 雷蛇 S2 反击电弧, 号角 S2 暴风号令
and S3 终极防线, 灰毫 S1 攻击力强化·γ型 and S2 专注轰击 take the basic strategy (DEFAULT) instead of the 下半 TANK class
row's TAKE_DAMAGE. Every other MANUAL 重装 skill, 深巡 S1 included, still waits for a hit.
- tools/build-data.mjs TRIGGER_DEVIATIONS: a small table keyed by the normal chess id and the skill id (the elite
follows through baseId), never by the skill id alone - 灰毫 S1 is the generic skcom_atk_up[3]. resolveTrigger applies
it last; rawRule keeps the official TAKE_DAMAGE; validateAll fails the build when an entry no longer meets a
TAKE_DAMAGE row. data/chess.json rebuilt with --offline: exactly the 12 records' rule changes.
- PR #12's kit lines (trigger 'DEFAULT' on 深巡 / 雷蛇 S2) stay and agree with the data; their comments and tests now
give this reason instead of the community summary.
- Tests: test/sim/feedback1-tank-triggers.test.js (the data, kit = data, a real battle on 战场#01(下半) in which each of
the six casts with an enemy in range before any hit and each of the other 46 MANUAL 重装 records casts on the first
hit); data.test.js trigger expectations; the 灰毫 / 号角 kit tests follow; docs-consistency.
- Docs: DESIGN §21.29 (new), §5.6, §19.2, §20.2, §21.20 (settled row), §21.23; DATA, SIM, PLAYING, CHANGELOG 0.1.1;
research 03 §1.4 drops the arknights.wiki.gg "offensive / defensive" summary and records the activity_table history
(上半 had no TANK row, 下半 added it) and this deviation; research 00-INDEX.
Conflicts resolved keeping both sides:
- effectsMeta.js header: the QA residual's "every [holder, item] pair, taken before the first item runs" and "an item
moved during it runs at most once", with the cell branch's wording of what 突变细胞 does mid-walk.
- DESIGN §6.6: the band pick that skips a strategy tied to a switched-off bond and the deployment of the operator a
transformation gains / the bench shed that spares it.
- DESIGN §21.25: PR #2's bench rule credited (the cell branch) plus the QA residual's snapshot-timing sentence.
- PLAYING: the new 突变细胞 sentence with the residual's hammer sentence (revives, borrowed hammers, the lend window).
- docs-consistency: both new tests (batch 6 QA residuals; 突变细胞 after the WA merge).
The band -> bond derivation moves out of GameData into shared/bandBonds.js,
which tools/build-data.mjs runs per band to write bands.json bondIds
(regenerated with --offline; 9 bands are tied, the build checks the ids).
GameData.bandBondIds reads the field, so the server and the client share
one source. DATA.md documents the field; the test checks the helper on
synthetic bands and that the data matches it.
--band makes the forced band the only one on offer; botPickBand would now
skip it for the default band in 标准 (潘格尼尼, 克莱门莎, 玛恩纳), so the
tool also clears the band's bond tie.
Official gameplay footage (bilibili BV1vzyVBuEN9 at about 8:24, BV1Qkw1zMEoR at about 7:25) shows the
next prep with the carrier's tile empty, one more deployment left, and the returned equipment plus the
new tier+1 operator on the bench, deployed by hand afterwards. That is the PRTS 备注 read as written
("生效时，原干员销毁，获得一名高一阶的随机初始干员"): a destroy followed by a gain. The closed PR #2 made
the same claim from a playtest.
- PlayerState.transformChess: the carrier is destroyed wherever it stands (board tile freed, summons
removed, copies back to the pool); its equipment, the cell included, returns to the hand first
(overflow temp, auto-merge); then the new operator is gained through acquireChess like any gained
operator (hand, overflow temp, the no-room rule of every gain; no summon card until it is deployed).
The returnItems option is gone (the cell's built-in was its only caller).
- A merge the gain completes merges like any gained copy: the elite takes a consumed deployed copy's
tile, else the hand. _mergeChess loses fromKey / fromDir, so the destroyed carrier's tile never
counts (PR #2 had kept that path; not adopted); audit.js drops its exception for it.
- builtinMeta, ctx.transform, items/meta.js and bot.js comments follow. Bots and AI 托管 deploy the
gained operator in their normal lineup step (no bot change needed; a real co-op round pins it).
- ITEM_RULES note and implFormula for the item card; data/items.json rebuilt with build-data --offline.
- Tests: facing, merge, playtest6-elite-to-board and items no longer pin the kept tile; feedback1-meta
covers a deployed and a bench carrier, hand full -> temp, hand and temp full, a merge with and
without a deployed copy (never the carrier's tile), 商业包装方案's count on the returned item, and bots /
AI 托管 redeploying; feedback1-cell checks the new note.
A rebuilt manifest leaves out every entry whose files are missing on that machine, so a run
with failed downloads silently dropped entries from the tracked file (PR #7 carried one 42
audio entries short; all 42 still resolve upstream). The run now keeps the current manifest,
prints the entries it would drop (also in the report) and exits 1. The tool runs main only as
a script so the guard and parseArgs can be tested; ASSETS.md documents the flag.
sganggs