بنقرة واحدة
sanmou-yanwu
يحتوي sanmou-yanwu على 5 من skills المجمعة من liamqma، مع تغطية مهنية على مستوى المستودع وصفحات skill داخل الموقع.
Skills في هذا المستودع
Updates `web/public/game-data/database.json` from the three 炎帝/UCL_Louis CSV source files (S<season>武将战法排行榜, S<season>影本战法, 阵容排行榜S<season>更新). Applies hero label/rank updates, skill note updates, and team updates while preserving the existing schema.
Produces a round-by-round (回合) analysis of a finished battle from a battle_log.txt (the OCR'd 战报详情 log), grounded in web/public/game-data/formula.md (the damage formula) and the skill descriptions in web/public/game-data/database.json. For each round it surfaces every hero's 减少伤害/增加伤害 (区间A/B/C multipliers), healing, and damage above a threshold, identifies the turning points (who killed whom and via which skill), explains the win/loss/draw cause, and can suggest a concise team/formation adjustment. Triggered when the user asks to analyse / 复盘 / 分析 a battle log and points at a battle_log.txt location.
End-to-end pipeline that turns battle-report screenshots on a USB-connected Android phone into a single de-duplicated, side-tagged, database-corrected battle log. Stage 1 pulls screenshots over ADB (no cloud round-trip): battle_detail_*.png (autojs/battle-detail.js) for ONE battle goes to study-battle-report/battles/<id>/images, and screenshot_*.png (native) goes to ./data/images. Stage 2 OCRs the scrolling 战报详情 frames for a battle into study-battle-report/battles/<id>/battle_log.txt, colour-tagging names blue=我方 / red=敌方 and snapping hero/skill/formation/bond names to web/public/game-data/database.json. Triggered when the user asks to pull/copy/import battle screenshots from the phone, and/or to OCR / scan / extract / read the battle screenshots into text.
Estimate a 战法's per-round strength and write *Estimate fields onto its entry in web/public/game-data/database.json, organised into the categories damage/healing/attribute/damageBoost/damageReduction/damageDealtReduction/damageTakenIncrease/evasion/lifesteal/crit/critDamage. Each estimate is a rough per-round coefficient (max-level, average-trigger) so the team-builder prompt can compare skills. Use when the user asks to "estimate" / "估算" / "evaluate the strength of" a named 战法.
Copies battle-report screenshots from a USB-connected Android phone (in /sdcard/Pictures/Screenshots/) directly to a local folder via ADB, with no cloud round-trip. Two filename patterns are supported - battle_detail_*.png (saved by autojs/battle-detail.js) goes to a per-battle dir study-battle-report/battles/<id>/images (default staging: battles/_incoming/images), and screenshot_*.png (native phone screenshots) goes to ./data/images. Triggered when the user asks to pull/copy/import battle screenshots from the phone.