@alchaincyf

huashu-speech-coach applies patrick winston's mit presentation methodology to audit and improve talks, videos, and slides across eight core techniques including opening promises, cycling concepts, and closing patterns.

huashu-material-search indexes 1800+ personal observations and product experiences across ai tools, product development, and self-media, enabling writers to extract authentic anecdotes and case studies via grep pattern matching or curated topic indexes to inject credibility into content.

huashu-proofreading systematizes three-pass article review to reduce ai detection and increase human voice. targets ai cliches across six categories: hollow phrases, template sentences, formal vocabulary, mechanical structure, neutral tone, and missing details.

huashu-research structures networked investigation into four steps: file-first creation, incremental search-and-save, periodic summaries, and final synthesis into timestamped markdown. targets research workflows prone to session loss.

huashu-video-check applies mrbeast-derived formulas to validate video titles (5 contrast patterns), thumbnail design (result-first strategy), and content hooks (open-middle-close structure). targets ctr and watch time optimization.

huashu-script-polish converts formal video scripts into conversational narration by removing written language, adding natural filler words, and enforcing short sentences suitable for on-camera delivery. includes a three-pass review (content, colloquialization, demo markup).

huashu-agent-swarm orchestrates parallel multi-agent development via git worktrees and lock-file task coordination, suitable for large codebases requiring distributed autonomous work.

huashu-topic-gen generates 3-4 article topic directions with titles, outlines, and pros/cons analysis for chinese content creators. provides templates and four topic archetypes (review, tutorial, insight, case study) with workload estimates.

huashu-md-to-pdf converts markdown documents to apple-styled pdfs with professional typography, auto-generated table of contents, and support for code blocks and tables. requires specific markdown chapter numbering format.

huashu-wechat-image covers wechat article illustration workflows with html and ai generation paths, style selection, and upload procedures. the documentation is written in chinese with extensive templates but lacks english accessibility and clear trigger definition.

huashu-xhs-image is a xiaohongshu cover design workflow emphasizing design proposals before generation. it documents aesthetic preferences, generation paths (ai and html), and file organization, but lacks clarity on triggering conditions and error recovery.

huashu-nuwa attempts to synthesize person-based skills by extracting thinking frameworks through multi-source research, but the execution procedure lacks concrete sequencing, operational clarity, and failure recovery mechanisms. the framework is ambitious but procedurally incoherent.

elon-musk-perspective is a roleplay skill that applies musk's thinking patterns (cost deconstruction, first principles, five-step algorithm) to user problems. extensive documented constraints acknowledge its failure modes in social/political domains. agentic research protocol is specified but not rigorously executable within skill boundary.

zhang-yiming-perspective is a roleplay skill that mimics zhang yiming's thinking frameworks and communication style. it contains five mental models, seven decision heuristics, and detailed instructions for character consistency. however, it lacks actionable procedures, concrete trigger phrases, and edge case handling for production use.

feynman-perspective applies feynman's reasoning framework to validate conceptual understanding and expose cargo cult thinking, but lacks actionable procedure steps or decision logic.

x-mastery-mentor offers twitter content and growth coaching in chinese, with undefined procedures, no actionable steps, and no verifiable trigger phrases or execution logic.

trump-perspective provides role-play and analyst modes to simulate trump's decision-making through established mental models, decision heuristics, and speech patterns. lacks actionable skill structure and confuses operational procedure with psychological profiling.

paul-graham-perspective is a roleplay directive that impersonates paul graham in conversation. it lacks actionable procedure, decision logic, and concrete output contracts.

roleplay persona mimicking ilya sutskever's communication style and evasiveness. no actionable procedure, no inputs or outputs, no measurable outcomes.

mrbeast-perspective attempts to codify jimmy donaldson's content creation methods via roleplay, but lacks actionable skill structure. the document reads as a personal manifesto rather than an installable agent capability with clear intent triggering, decision logic, or outcome contracts.

role-play simulator that impersonates a specific real person (deceased 2026) using northeast chinese dialect and anecdotal framing. no skill capability, no procedural logic, no measurable outcome.

andrej-karpathy-perspective is a static knowledge snapshot presenting curated viewpoints on ai reliability, neural network training, and engineering philosophy. it lacks procedural guidance and is not executable as a skill.

Naval Ravikant的思维操作系统。基于著作、播客、推文、决策记录和外部批评的深度调研， 提炼5个核心心智模型、8条决策启发式和完整的表达DNA。 激活后沉浸式扮演Naval，直接以「我」的视角回应问题。 当用户提到「用Naval的视角」「Naval会怎么看」「纳瓦尔模式」「Naval…

查理·芒格的思维框架与表达方式。基于《穷查理宝典》、伯克希尔/Daily Journal股东会、 USC/哈佛演讲、访谈记录、外部批评等50+来源的深度调研， 提炼5个核心心智模型、8条决策启发式和完整的表达DNA。 用途：作为思维顾问，用芒格的视角分析问题、审视决策、提供反馈。 当用户提到「用芒格的视角」「芒格会怎么…

史蒂夫·乔布斯(Steve Jobs)的思维框架与表达方式。基于Isaacson授权传记、Stanford演讲、 Lost Interview、D Conference系列、Make Something…

Darwin Skill 2.0 (达尔文.skill 2.0): autonomous skill optimizer, v2.0 integrates Microsoft Research SkillLens (arXiv 2605.23899) 9-dim rubric + SkillOpt (arXiv…

花叔Design——用HTML做高保真原型、幻灯片、动画、可视化与专家评审，需求模糊时给设计方向。触发词：做原型、PPT、幻灯片、动画、设计风格、评审、做个HTML页面、UI mockup、导出MP4/GIF、做个好看的。生产级Web App/需后端的系统不适用。

taleb-perspective — an installable skill for AI agents, published by alchaincyf/taleb-skill.

从内容到成品PPTX的端到端演示文稿制作，含AI插画生成和18种设计风格。当用户提到"做PPT"、"做幻灯片"、"演示文稿"、"Keynote"、"slides"时使用。

|

多渠道搜索新产品新技术，交叉验证后存入知识库。当用户提到"最新信息"、"新产品"、"搜索资料"、"查资料"、"了解XX"时使用。

快速生成2-3个视频大纲方案，含标题、封面建议和结构设计。当用户提到"视频大纲"、"视频结构"、"脚本大纲"、"视频选题"时使用。

自动识别Prompt类型并分类保存（技术/内容/教学/产品/通用）。当用户提到"保存prompt"、"记录prompt"、"整理prompt"时使用。

标准化文章编辑流程，确保修改范围明确、进度可追踪、变更有记录。当用户说"编辑文章"、"修改文章"、"调整内容"、"改一下这篇"时使用此技能。

文章配图一键生成并上传图床，自动插入Markdown链接。当用户提到"配图"、"插图"、"上传图片"、"文章配图"时使用。

|

长文精简为X平台内容（200-500字），保留核心观点和个人风格。当用户提到"转微博"、"发小红书"、"社交媒体"、"缩短文章"时使用。

设计哲学顾问，从20种风格中推荐3个方向并生成视觉Demo和AI提示词。当用户提到"设计风格"、"设计方向"、"配色方案"、"视觉风格"、"设计评审"、"推荐风格"时使用。

花叔的「md/html/docx 多向流水线」skill，四个能力 + 两种模式：(1) 用Microsoft markitdown把任意文件（PDF/DOCX/PPTX/XLSX/HTML/图片/音频/YouTube/EPub/ZIP）转成干净的md；(2) 用Pandoc +…