dososo

Use when explicitly turning a repeated workflow, SOP, prompt, document, expert process, or tool routine into a reusable Agent Skill package, or improving, validating, packaging, publishing, or governing an existing Claude/Codex/Agent Skill; especially for SKILL.md, references/scripts/assets, evals, gotchas, route tests, Non-Skill decisions, and lifecycle governance.

将中文长文章、作者头像、个人照片、产品截图、应用界面截图或粗略观点转成小红书优先的 6 张竖版图文卡片创作流程。用于用户要求产出小红书/RED 轮播图、竖版 3:4 图文卡、中文知识卡片、产品观点卡片、图片生成提示词、设计稿说明或可交付的六页卡片脚本时。

Use when converting Chinese long-form articles, author/avatar images, and product screenshots into a Xiaohongshu-first six-card vertical carousel plan, storyboard, visual brief, image prompts, or QA checklist; especially for 小红书图文卡片、6张竖版卡片、长文视觉化、头像合成、产品截图种草、中文内容卡片.

审视、设计、从零启动或改进 HarmonyOS、OpenHarmony 与 ArkUI 产品界面。用于检查 ArkTS UI、建立新项目设计基线、区分永恒原则与当前平台视觉语言、规划跨手机/平板/折叠屏/PC 的导航和布局、多输入控件状态、animateTo、transition、geometryTransition、springMotion、手势跟手与可打断动效、异步状态真实性、视觉分层参数、无障碍或动画性能，或建立鸿蒙 Motion Token 时。不要用于签名、构建、网络、存储、权限、数据库等纯工程问题，也不要用于通用视频剪辑或其他移动平台的设计规范评审。

把用户对 HarmonyOS、ArkUI、Canvas 或代码生成产品动效的模糊描述转换为准确术语、相近概念和可能的 ArkUI 能力，例如“松手像急刹车”“拖过头有阻力”“页面像从同一个卡片长出来”“后端没完成却显示成功”。用于命名、检索和编写提示词；不要用于完整设计审视、传统视频剪辑、镜头叙事、直接实现或构建问题。

严格审视 HarmonyOS、OpenHarmony 与 ArkUI 产品或代码，只报告有证据的设计、导航、跨设备适配、输入状态、动效、异步状态真实性、无障碍和动画性能问题，并给出规则 ID、ArkUI 落点与通过结论。用于用户明确要求 review、audit、审查、验收或评估鸿蒙 UI；不要用于从零设计、直接实现功能、修复构建签名或审查非界面代码。

当用户需要把内容（文章、笔记、截图、数据、产品说明）做成中文图文卡片组图时使用——小红书 / Rednote 3:4 竖图、公众号封面、社交卡。提供三套独家视觉语言（静纸 Still Paper本手记 / Signal Proof 证据层 / Bridge Canvas 跨平台桥），由 Claude 读懂内容、自己撰写 brief.json，交给确定性引擎渲染出第一眼惊艳、一眼可辨、国内开箱即用的图文组图。

Use this skill when the user wants to turn an article, lesson, product note, retrospective, method, or abstract idea into JUJU-style hand-drawn visual method images. It analyzes the content, asks for the output ratio, decides whether to create one image or a series, writes image-generation prompts, and calls the available image-generation tool when possible.

把一个目标/主题/文章/截图/数据转成有观点、有节奏、有视觉约束的单文件 HTML 演示（网页 PPT、科技战报、产品拆解、行业观察、路演/发布会 deck、社媒可传播长图）。**7 套各锚定公认设计体系的视觉人格**——技术崇高派 Technical Sublime（工程冷光·技术战报/算力）/ 构成主义 Constructivism（宣言/号召）/ 信息·数据设计（Tufte·数据/图表）/ 判断震慑极简 Luxury（高端品牌/判断）/ De Stijl 新造型（标准/秩序）/ 编辑主义 Editorial（衬线·深度观察）/ 演示禅 Zen（暗墨·思想/品牌 keynote）——用户可指定，否则按内容自动判选并透明告知。token + validator 机检（P0/P1/P2）+ 32 维审计**机器强制设计一致性**；反伪造（不造数据/logo/截图/来源·示意标注）；字体本地子集；图按角色路由（默认 CSS/SVG·要图先探测生图环境）。Generates single-file HTML decks with 7 design-system-anchored visual personas, a semantic layout grammar, locked design tokens, evidence-grading, machine validation (P0/P1/P2 + 32-dimension audit), an anti-fabrication discipline, and image-generation environment detection. Use when the user wants a presentation / deck / slides / 网页 PPT / 战报 / 产品拆解 / 行业观察 / 路演 / 发布会 / 社媒长图, or asks to turn an article / screenshot / dataset into slides. Do NOT use for large data-table training material, strict legal/medical/financial compliance documents, multi-author collaborative PPT editing, plain-te

BLCaptain 学习工作台。Use when the user wants to learn a topic or skill over multiple turns/sessions, especially with a concrete deadline, deliverable, exam, work project, skill-building goal, or request such as "教我", "我想学习", "带我学", "做一个学习计划", "每天推进", "复盘学习", or "把资料变成课程". Maintains an evidence-gated learning workspace with learning briefs, curated sources, micro-lessons, reusable cards, progress records, terminology, reviews, and next-step plans adapted to concrete goals, real deadlines, source constraints, and reusable learning artifacts.

Build and run long-running AI agent workflows for Codex, Claude Code, Cursor, and other coding agents with contracts, disk-backed state, role separation, verification traces, restart criteria, subjective scoring, and harness cleanup. Use when a user asks an agent to complete multi-step coding, product, research, content, design, debugging, batch generation, or cross-session tasks that require planning, repeated execution, evidence, or recovery after failures.

Improves research quality when accuracy matters; work depends on external facts such as documentation, APIs, pricing, licenses, standards, current events, research, or source code; a question is complex or multi-part; confidence is low; or sources conflict. Excludes casual chat, pure formatting, mechanical edits, and trivial facts already known with high certainty.