Agent skill
podcast-script-generator
Transform structured podcast notes into engaging video scripts with a conversational, first-person broadcasting style. Use when user asks to "generate podcast script", "convert to broadcast style", "make video script", "rewrite for video", "口播脚本", "改成口播", or wants to turn podcast summaries into presenter-style content for short videos.
Install this agent skill to your Project
npx add-skill https://github.com/zephyrwang6/myskill/tree/main/podcast-script-generator
SKILL.md
Podcast Script Generator
Transform structured podcast notes into engaging, presenter-style video scripts.
When to Use
Use this skill when the user wants to:
- Convert structured notes into presenter-style video scripts
- Rewrite podcast summaries for short video content
- Create conversational broadcast content from formal notes
- Generate scripts with strong personal voice and recommendations
Workflow
Step 1: Get Source Material
Ask the user for the content source:
- Direct paste: User pastes text directly
- Daily podcast file: File from
/Users/ugreen/Documents/obsidian/每日播客/ - Other file path: Any other markdown file
If user provides a file path, read it with Read tool.
Step 2: Analyze Content Structure
Identify:
- Title & Guest: Podcast name and main speaker
- Core Theme: Main topic discussed
- Key Points: Numbered list of insights (usually 3-12 points)
- Highlights: Dialogue excerpts or quotes
- Data/Examples: Statistics, case studies, comparisons
Step 3: Apply Broadcasting Style
Read references/style-reference.md and references/example-conversions.md to internalize the target style.
Opening (Varied, Natural)
Avoid the fixed template. Instead, choose from these natural openings based on content:
- "今天要给你推荐一期播客..."
- "最近听到一期特别有意思的节目..."
- "这期是我最近听到的对[领域]非共识观点最多的一期..."
- "这个月听到的最有收获的一期播客..."
- "如果你关注[话题],这期播客你一定要听..."
Then introduce the guest naturally:
- "嘉宾是[名字],[一句话身份]..."
- "介绍一下[名字],[成就/背景]..."
Add a hook specific to the content:
- "她几个月前演讲说'不要相信设计流程',3个月后自己就觉得过时了..."
- "他可能是华尔街唯一一个既能看懂财务报表,又能看懂火箭发动机的人..."
Content Transformation Rules
Rule 1: No "发言人" markers
- Direct output without speaker labels
- Use paragraph breaks to indicate natural pauses
Rule 2: Banned sentence patterns
DO NOT use these rigid contrasting patterns:
- ❌ "不是...而是..."
- ❌ "是...不是..."
- ❌ "被...倒逼着..."
- ❌ "从...变成...,从...变成..." (overly structured parallel)
Instead, express contrast naturally:
- ✅ "设计不是主动想变,是工程端的变化逼着你必须变"
- ✅ "以前大家把这套流程当真理,但现在基本已经死了"
- ✅ "工程师已经转向命令行和agent模式,而设计师开始用IDE了"
Rule 3: Use conversational connectors SPARINGLY
Limit these phrases (max 1-2 per section):
- "说白了" → use "意思是" or just omit
- "这什么意思呢" → use "为什么?" or just explain directly
- "所以你看" → use "结果就是" or just show the result
Rule 4: Natural transitions
Instead of rigid structure markers, use:
- "第一个让我意外的点是..."
- "还有一个特别反直觉的发现..."
- "这里有个特别有意思的地方..."
- "更夸张的是..."
- "说到这个,就不得不提..."
Rule 5: Personal commentary
Insert reactions naturally:
- "我觉得这个点特别值得思考"
- "特别反直觉"
- "特别有意思"
- "让我特别意外"
Rule 6: Explain concepts conversationally
When explaining complex ideas:
- Start with the problem or context
- Walk through the logic step by step
- Use "打个比方" for analogies when needed
- End with "就是说" or just the conclusion
Closing
End naturally:
- "好了,今天就聊到这..."
- "今天就分享到这里..."
- "今天就聊到这,如果你也[相关话题],强烈建议去听听完整播客..."
- "我是[昵称],下次见"
Step 4: Output Format
Produce:
- Full script without speaker markers, with natural paragraph breaks
- Estimated duration (250 words ≈ 1 minute)
- 3-5 suggested titles for the video
- Recommended hashtags/topics
Style Guidelines
Paragraph Structure
- Each paragraph: 2-4 short sentences
- Use empty lines between paragraphs for pacing
- One idea per paragraph
- Key data/quotes can stand alone
Tone
- Conversational, like talking to a friend
- First person "我" perspective
- Natural enthusiasm without exaggeration
- Occasional rhetorical questions
Language
- Avoid written-style transitions
- Prefer short sentences
- Use spoken Chinese patterns
- Keep technical terms but explain them naturally
Reference Materials
references/style-reference.md- Full example of target stylereferences/example-conversions.md- Before/after conversion examples with techniques
Constraints
- Preserve all factual information - change style, not substance
- Maintain accuracy of data, quotes, and attributions
- Target length: 750-1500 words (3-6 minutes of video)
- If source is too long, ask user which points to prioritize
- NO "发言人" markers
- NO rigid "不是...而是" patterns
- Use "说白了" max 1-2 times in entire script
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
start-work
每日工作启动助手。读取Obsidian收件箱、计划文件,提醒今日待办,询问内容创作计划,展示周计划进度,调用热点采集。触发词:"开始工作"、"开启新一天"、"今天做什么"。帮助用户快速进入高效工作状态。
doc-coauthoring
Guide users through a structured workflow for co-authoring documentation, articles, or long-form content. Use when user wants to write documentation, proposals, articles, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting articles, or using "co-authoring" workflow.
mem-query
AI个人记忆系统的记忆查询功能。检索各层级记忆文件,综合多层级信息回答用户问题。使用场景:(1) 用户问"我的记忆中关于XXX"时;(2) 用户询问自己的习惯/偏好/价值观时;(3) 需要基于用户历史提供建议时。该skill会自动检索L1-L4各层级,引用来源,给出基于记忆的个性化回答。
article-review
根据原文内容撰写深度文章评价/解读。当用户提供一篇文章、博客、公众号文章或任何长文内容,并要求生成评价、解读、读后感或二次创作内容时使用此技能。适用于:(1) 对技术文章、行业分析、年终总结等进行深度解读,(2) 提炼文章核心观点并用通俗语言重新表达,(3) 为社交媒体传播生成二次内容。
mem-record
AI个人记忆系统的记忆记录功能。自动从对话中提炼关键信息并记录到相应层级。使用场景:(1) 用户说"记录到记忆系统"、"记住这个"、"把这次对话记下来"时;(2) 检测到重要事件、决策、偏好表达时;(3) 用户完成重要任务或做出决策时。该skill会自动判断应该记录到L1情境层、L2行为层、L3认知层,还是建议更新L4核心层。
remotion-video
使用 Remotion 框架以编程方式创建视频。Remotion 让你用 React 组件定义视频内容,支持动画、字幕、音乐可视化、3D 视频、教程讲解视频等。适用于程序化视频、批量生成、数据驱动视频、音乐可视化、自动字幕等场景。
Didn't find tool you were looking for?