Coding & Dev Tools

ariname-lilith Bundle

从卡拉彼丘 BWIKI 角色主页面和画廊查找图片，按角色名、立绘关键词、尺寸、纵横比、格式及排除词评分。用户要求寻找角色立绘、全身图、默认外观或为视觉分析选择图片时使用。

ariname-lilith Bundle

自动发现并提取卡拉彼丘 BWIKI 中角色的誓约、档案馆、小传、剧情、羁绊、语音、通讯、回归信和生日剧情。用户要求深挖人物经历、性格、关系、语言风格或角色扮演素材时使用。

ariname-lilith Bundle

将卡拉彼丘剧情资料整理成目标角色的有来源事件时间线，支持明确日期、相对顺序、参与者、影响、事实与推断标记。用户要求梳理经历顺序、去重事件或检查剧情前后矛盾时使用。

ariname-lilith Bundle

归一化卡拉彼丘 BWIKI 的姓名、性别、阵营、生日、星座、年龄、身高、体重、活动区域、兴趣爱好和饮食习惯，并为候选值保留页面级来源。用户要求建立角色卡基础身份事实表或检查资料冲突时使用。

ariname-lilith Bundle

综合卡拉彼丘主档案、誓约小传、剧情、通讯与语音等全部已发现资料，建立有原文证据但不照抄来源的经历摘要和可执行第三人称人格模型。用户要求生成角色卡背景、角色灵魂、行为逻辑或核查人物分析时使用。

ariname-lilith Bundle

强制实际打开卡拉彼丘默认立绘并记录发型与发色、眼睛、面部、体型观感、日常服装、饰品及整体气质，附视觉置信度和图片来源。用户要求生成角色卡外貌或核对立绘时使用。

ariname-lilith Bundle

从卡拉彼丘主档案、誓约、剧情、通讯和语音等全部资料中综合提取有原文证据的角色、组织与群体关系，并生成单卡初始关系或全角色双向关系总表。用户要求分析人物关系、补充角色扮演互动起点或生成关系表时使用。

ariname-lilith Bundle

---

lqyeyao6666 Skill

查看并计算缓存命中率。用户提到缓存命中率、cache hit rate、命中次数、未命中次数、缓存效率或想知道缓存表现时使用；适用于代码库、日志、监控导出和运行时指标，优先读取现有数据，不凭空创建或估算指标。

610551763-maker Bundle

总控自媒体视频稿件拆分镜：逐字保留原稿，以一个完整事件、因果链、观点或稳定画面任务为分段边界，不按句号机械拆行；以链接真实可播放和画面质量为首要标准，检索剪辑素材、精确同期声、动画参考和证据来源，最终生成逐段一一对应、可直接打开的可视化分镜。

610551763-maker Bundle

为已经拆分并锁定画面方案的自媒体视频分镜核验技术原理、模型架构、研究结论、报道、公告、原始文件、人物原话和数据来源，提供可直接核对的官方或一手链接；将事实与原理依据同动画画面参考明确分层，仅在动画片段确有数据关系且方案明确采用图表时生成数据卡 PNG 和来源链接。用于单独补原理/报道/数据来源/数据卡，也由 storyboard-breakdown 在完整拆分镜中默认调用。不得改稿、决定剪辑或动画、替代视频素材或画面参考检索，或直接写入云文档。

610551763-maker Bundle

为已完成拆分的剪辑片段检索可直接播放的高清 B-roll、发布会、采访和新闻档案；以链接有效、内容准确、画面干净和原始上传为先，再按国内/海外媒体生态分配平台权重。所有视频都必须给实际核准的精确起止时间和该段画面；同期声还须给原话、翻译和对应原稿。

610551763-maker Bundle

为已经完成拆分、明确归类为动画且锁定了信息关系与简明视觉表达的自媒体视频分镜，按信息结构、构图排版、运动衔接、材质风格等目标借鉴维度分别检索外部视觉作品；优先使用动画结构词而非稿件中的公司名、行业名或单个主题词，每个必需维度实际打开并比较不少于 3 个候选，再按结构匹配、直接映射、制作可执行性、来源质量和可见热度排序。用于单独补画面/动效参考，也由 storyboard-breakdown 默认调用。不得决定动画方案、重新拆稿、把原理或事实页面冒充画面参考、生成图片冒充参考或直接写入云文档。

hajekt2 Bundle

Check Bitwarden access, create generated secrets, and run approved credential-consuming operations without returning values. Use for bws, Bitwarden-backed deployment, signing-key import, or OpenTofu state inventory.

mwpryer Bundle

Read and write Firestore through firesmith, a schema-first typed wrapper over both Firestore SDKs (firebase-admin and firebase web) from one Standard Schema definition. Use when a project imports "firesmith", when writing Firestore data-access code in such a project, or when the user mentions firesmith, typed Firestore collections, or schema-typed Firestore queries.

waldfalke Bundle

Разрешение инженерных/изобретательских противоречий LT-методом Бушуева–Бартини–Кузнецова: факторам противоречия назначаются физические размерности LᵐTⁿ (таблица Бартини), перемножаются по логике «И-И», по тренду вещественно-полевых ресурсов (ВПР) опознаётся икс-элемент. Главное — метод указывает не одно решение, а целый тренд физически-эквивалентных рычагов, среди которых есть неочевидные. Триггеры: «реши по Бартини», «LT-метод», «таблица Бартини», «метод И-И», «противоречие → размерность → икс-элемент», «найди неочевидное/контринтуитивное решение через размерности».

haumer Skill

Gemini

haumer Skill

Annotated live product walkthrough in the user's browser — impersonate a fresh (or named) account, drive a user journey step by step with an on-screen narration bar, support replaying steps, and collect every bug hit along the way. Use when the user says "show me X as a customer", "walk through Y", "do the steps", or names a persona/journey to demo.

lbachelotcapitalb Bundle

Préparer une contribution propre et adoptable à un repo dont tu n'es PAS mainteneur, de bout en bout : resync du fork, branche courte, respect des règles immuables et des conventions du projet, garde de validation (tests/lint/build), puis PR petite et prête à relire. Utilise ce skill quand tu commences à coder une feature ou un fix sur le repo de quelqu'un d'autre. Pendant « contributeur » du skill review-contrib (côté relecteur). Version exécutable d'un bon CONTRIBUTING.md.

enrikkk Skill

Re-explain your own immediately previous message in plain, simple language with no jargon, assuming the user does not already know the technical background. Use whenever the user says they did not understand, that you were too technical, that you were spitting jargon, asks you to explain it simpler or more clearly, or invokes /drop-jargon explicitly.

enrikkk Skill

Explain a function, class, or file block-by-block in plain English. Use whenever the user asks to explain, walk through, break down, or understand what a piece of code does (e.g. "explain this function", "walk me through X", "what does this file do"), or when invoked explicitly as /explain-code <path or symbol>.

ricardooschutz Bundle

Use when auditing, reviewing or designing a UI (pantalla, módulo, formulario, tabla, dashboard, flujo, drawer, navegación) and the user asks for puntos débiles, simplicidad, facilidad de elección, optimización, fricción, carga cognitiva, or names a UX law (Hick, Fitts, Jakob, Miller, Doherty, Von Restorff, Tesler, Postel, Occam, Pareto, Gestalt, peak-end, Zeigarnik). Also for /ux-laws.

gyy489 Skill

Start, stop, or check the operator's local Skills/Capabilities relationship dashboard through the Two-Headed-Wu skills-dashboard capability. Use when the user wants to browse or visualize how Skills relate to each other (skill_set membership, capability ownership, runtime availability, or possible cross-references) in a web page, or asks whether such a dashboard is running. It is on-request only, not a standing service: start it when the user asks to view it, and either let its idle timeout close it or stop it explicitly once they say they're done. Never use it to expose, proxy, or bind this server to anything other than loopback, and never treat its "inferred" edges as verified dependencies.

gyy489 Bundle

Query the operator's redacted four-machine/two-site map and, within the user's installed limited local publication scope, let Codex publish checked static output to the fixed private-site or public-site site through a protected local IPC executor. Use when Codex or Claude Code needs to inspect these registered resources or receives a request such as “把这个网页部署到临时公开页面”. Do not use for generic SSH, arbitrary hosts, FRP, Nginx, HTTPS configuration, DNS, databases, passwords, private records, or dynamic-service deployment.

gyy489 Skill

Bootstrap or inspect the operator's fixed Aliyun Two-Headed-Wu runtime area after an explicit owner request, including SSH reachability, protected authentication, the shared non-login system identity, and pre/post guards for the two registered sites. Never use for arbitrary SSH, commands, hosts, paths, website changes, or credential inspection.

amart-builder Bundle

Session-wide CEO mode led by GPT-5.6 Sol. Sol keeps judgment, synthesis, user-facing authorship, and its own coding implementation; treats Claude Opus 5 as the cross-family peer in both seats it holds - a read-only adviser during planning, and an independent fresh-context adversarial reviewer of every substantive result before completion is claimed; and delegates bounded work to GPT-5.6 Terra/Luna, native Codex subagents, Claude workers, or Grok. Sol makes every final decision. Use when the user invokes $orchestrator, says orchestrator mode, delegate mode, manager mode, or asks GPT-5.6 to lead a team. Keep the mode active until the user says orchestrator off.

sbradl Skill

Red phase of TDD — one failing test, proven to fail for the right reason. Use when the user wants a failing test first, or when tdd-loop needs the red phase.

sbradl Skill

Runs the test-driven loop — drives tdd-red and tdd-green one slice at a time and, at each batch boundary, a craft-review-fed tdd-refactor pass. Use when the user wants to build a feature or fix a bug test-first, or says red-green-refactor.

sbradl Skill

Green phase of TDD — the smallest source change that makes a failing test pass. Use when the user has a red test and wants it green, or when tdd-loop needs the green phase.

sbradl Skill

Refactor phase of TDD — improve green code's design without changing behaviour or any assertion. Use when the user asks to refactor, names a smell, or acts on a review finding, or when tdd-loop needs the refactor phase.

kikakikakikakika Skill

Run this automatically before presenting ANY design output — UI screens, flows, components, wireframes, copy+layout combinations, or interaction specs. Review the planned design against Nielsen's 10 Usability Heuristics, fix all violations internally, then present only the resolved output. The person who requested the design should never see a version with heuristic violations.

kikakikakikakika Bundle

Apply or review visual style and brand guidelines. Always ask which brand before proceeding. Use when working on design, layout, visual assets, presentations, documents, or any visual output.

kikakikakikakika Skill

Head of Design only. Reads from design-review-monitor persistent storage, maps review data to 22 design competencies, and generates spider/radar charts per person and for the team overall. Run monthly.

zmjjkk123-hub Skill

Forge GUI 系统：MenuType 注册（MenuSupplier、IContainerFactory、IForgeMenuType#create）、AbstractContainerMenu 菜单（#stillValid、#quickMoveStack、ContainerLevelAccess）、数据同步（Slot、SlotItemHandler、DataSlot、ContainerData、SimpleContainerData）、菜单打开（ServerPlayer#openMenu、MenuProvider、SimpleMenuProvider）、屏幕系统（Screen、AbstractContainerScreen、Renderable、GuiEventListener、ContainerEventHandler、NarratableEntry）、GuiGraphics 渲染（#hLine/#vLine/#fill/#fillGradient、#drawString/#drawCenteredString、#blit 纹理渲染、ARGB 颜色）、相对坐标与 Blit Offset、Overlay 覆盖层（ForgeLayeredDraw、ForgeLayer、AddGuiOverlayLayersEvent、vanilla 渲染顺序）、屏幕组件（#init、#tick、#render、#renderBackground、#renderTooltip）、容器屏幕（imageWidth/imageHeight、leftPos/topPos、#renderBg/#renderLabels）、MenuScreens#register 注册、FMLClientSetupEvent 客户端设置、FeatureFlagSet、 FriendlyByteBuf 额外数据。

zmjjkk123-hub Skill

Forge 杂项功能：配置系统（ForgeConfigSpec、ForgeConfigSpec$Builder、ConfigValue、defineInRange/defineInList/defineList/defineEnum/define 方法、ModLoadingContext#registerConfig、Config.Type CLIENT/COMMON/SERVER、ModConfigEvent$Loading/Reloading 事件）、调试分析器（Debug Profiler、F3+L、ProfilerFiller#push/#pop）、GameTest 游戏测试（@GameTest、GameTestHelper、@GameTestGenerator、@BeforeBatch/@AfterBatch、@GameTestHolder、RegisterGameTestsEvent、TestFunction、模板结构、/test 命令 run/runall/runthis/runthese/runfailed）、按键绑定（KeyMapping、RegisterKeyMappingsEvent、InputConstants$Type/KEYSYM/SCANCODE/MOUSE、IKeyConflictContext、KeyModifier、ClientTickEvent 检查、ScreenEvent$KeyPressed/ScreenEvent$MouseButtonPressed）、Forge 更新检查（updateJSONURL、VersionChecker#getResult、IModInfo、ModContainer、版本状态 FAILED/UP_TO_DATE/AHEAD/OUTDATED/BETA_OUTDATED/BETA/PENDING）。

zmjjkk123-hub Skill

Minecraft Structure Set 结构集定义：STRUCTURE_SET 注册表、data/<namespace>/worldgen/structure_set/ 数据包路径、tags/worldgen/structure_set/ 标签、JSON 格式（structures 可放置结构特性列表 structure+weight、placement 放置行为 salt 随机种子盐 frequency 尝试概率 0-1 frequency_reduction_method 频率降低方法 default/legacy_type_1/legacy_type_2/legacy_type_3、exclusion_zone 排除区域 chunk_count+other_set、locate_offset /locate偏移 chunks -16..16、type 放置类型）、Placement Types 放置类型（concentric_rings 同心环 固定数量 环绕世界中心：distance 环宽度+间隙 6区块单位 0-1023、count 总尝试次数 1-4095、spread 中心环尝试次数 0-1023、preferred_biomes 首选生物群系；random_spread 随机扩散 分割维度为spacing区块单元格 每个尝试一次 结构仅在separation区块内生成：spread_type linear/triangular、spacing 平均距离 1-4096区块、separation 最小距离 0<spacing；dimension_origin 维度原点 26.3 放置在维度原点 噪声生成器首选spawn_target 否则区块(0,0)）、Definition Behavior 定义行为（服务器启动加载一次、/reload 不重新加载、生成期间每个维度评估可用集 结构仅在概率/生物群系/地形检查通过时放置）。