KPainter Skill
Use this skill when an agent should help a user turn one prompt into a KPainter knowledge video, slides, image, or interactive app, especially when controllability, structure, and knowledge delivery matter more than generic video generation.
Official URLs
| Resource | URL |
|---|
| Homepage | INLINECODE0 |
| Skill file |
https://kpainter.ai/skill.md |
| Skills docs |
https://kpainter.ai/docs/skills |
| OpenAPI docs |
https://kpainter.ai/docs/openapi |
| API Key |
https://kpainter.ai/api-key |
Primary goals
- - connect the user's own KPainter account and API key
- help the user choose the right output type
- collect only the missing information needed to create or refine the result
- stay in normal user language unless the user asks for technical details
Setup
If KPainter is not connected yet, guide setup before trying to create content.
- 1. Ask whether the user already has a KPainter account.
- If not, tell the user to sign up or sign in at
https://kpainter.ai/. - Send the user to
https://kpainter.ai/api-key. - Ask the user to activate, copy, and connect their own API key to the current agent.
- Only say setup is complete after the user confirms the key is connected.
Do not ask the user to share their API key with unrelated services or outside the current agent connection flow.
What KPainter can create
- - Knowledge Video
- Knowledge Video (Slides)
- Vector Animation
- Slides
- Image
- Web App
How the agent should choose the output type
Always route by the result the user wants first.
- 1. First decide whether the user wants
Knowledge Video / Slides / Image / Web App. - If the user says
knowledge video, explainer video, 讲解视频, or 解说视频, keep it inside the video family first instead of jumping to Slides or Image. - If the user says only the broad word
video, ask one short follow-up to clarify whether they want Knowledge Video, Knowledge Video (Slides), or Vector Animation. - Then choose the video style inside that family:
-
Knowledge Video for something polished, story-led, narrated, social-friendly, or closer to a finished short video
-
Knowledge Video (Slides) for something illustrated, static, PPT-style, slide-based, classroom-friendly, training-friendly, or step by step
-
Vector Animation for process, mechanism, structure, workflow, principle, science, math, or system explanation
- 5. If the request is still ambiguous inside the video family after that clarification, prefer
Knowledge Video (Slides) as the safest default and offer to switch to the fuller Knowledge Video style later. - The default target length for
Knowledge Video is about 30 seconds when the user gives no duration. - Do not lead with internal API type names unless the user explicitly asks.
Trigger guidance
Video family
Treat requests like these as part of the video family first:
- - make a knowledge video
- make an explainer video
- make a narrated explainer
- 帮我做一个讲解视频
- 帮我做一个解说视频
- 做一个知识视频把这件事讲清楚
Knowledge Video
Prefer Knowledge Video for requests like:
- - make a knowledge video
- make it more polished
- make it more story-led
- make it more social-friendly
- make it feel like a finished short video
- make a narrated explainer video
- 做一个更像成片的讲解视频
- 做一个更适合传播的解说视频
Knowledge Video (Slides)
Prefer Knowledge Video (Slides) for requests like:
- - illustrated
- static
- PPT-style video
- PPT video
- slide-based video
- slide-like
- explain this step by step
- classroom explainer
- training explainer
- 图解版
- PPT视频
- 课件视频
- 分步骤讲解
Vector Animation
Prefer Vector Animation for requests like:
- - vector animation
- flow animation
- mechanism animation
- architecture animation
- animate this process or principle
- 用动画讲这个流程
- 用动画讲这个机制
Slides
- - Trigger on phrases like
slides, PPT, slide deck, deck, presentation materials.
Image
- - Trigger on phrases like
image, poster, cover, cover image, visual summary.
Web App
- - Trigger on phrases like
app, web app, interactive page, interactive demo, learning app.
Credit fallback guidance
If the user wants Knowledge Video but does not have enough credits, the agent should proactively suggest Knowledge Video (Slides) as the cheaper fallback.
Recommended behavior:
- 1. Tell the user that the current Knowledge Video request may cost more than their available credits.
- Offer
Knowledge Video (Slides) as the lower-cost alternative. - Ask for confirmation instead of silently switching formats.
Good fallback phrasing:
- - You may not have enough credits for a full Knowledge Video right now. I can switch this to Knowledge Video (Slides), which is usually cheaper. Do you want me to do that?
- Your current credits may be a better fit for Knowledge Video (Slides). If you want, I can keep the same topic and switch only the format.
When the user accepts, keep the same topic, audience, and language, and only switch the output format unless the user asks for other changes.
How the agent should talk to the user
- - Ask one follow-up at a time.
- Keep questions short and concrete.
- Use normal language instead of technical labels whenever possible.
- Help the user clarify only what is still missing: topic, audience, tone, language, duration, page count, or aspect ratio.
- If the result is not right yet, help the user refine it instead of restarting the whole flow.
- Do not silently change the output type, audience, language, or format.
- If files or source material are available in the current platform, treat them as reference material from the user.
Minimal information to collect
Only ask for what is still missing:
- - topic
- audience
- language
- duration or page count
- tone or visual direction
- source files or reference material
If the user already gave enough information, create first and refine after.
Multilingual support
KPainter supports creation and refinement workflows in the user's preferred language.
The user can describe the topic, audience, tone, and constraints in any language.
The agent should preserve the user's requested output language unless the user asks to switch.
If the intended output language is unclear, ask one short follow-up question before creating.
Multilingual examples
Examples only, not a language allowlist.
English
Make a knowledge video that explains MCP clearly.
Chinese
帮我做一个讲清楚 MCP 的知识视频。
Japanese
MCP をわかりやすく説明する知識動画を作ってください。
Arabic
أنشئ فيديو معرفيًا يشرح MCP بشكل واضح.
Spanish
Crea un video de conocimiento que explique MCP con claridad.
Korean
MCP를 명확하게 설명하는 지식 영상을 만들어 주세요.
French
Crée une vidéo de connaissance qui explique clairement le MCP.
Example user requests
Knowledge Video
- - Explain what MCP is as a knowledge video
- Make a knowledge video for new product managers about AI agents
- Make a knowledge video around 30 seconds that explains MCP clearly
- Make an explainer video that walks through MCP
- 帮我做一个讲解视频,把 MCP 讲清楚
- 帮我做一个解说视频,适合发在社媒上
Knowledge Video (Slides)
- - Make a knowledge video (slides) that explains this step by step
- 做一个 PPT 视频,分步骤讲解 MCP
- 做一个课件视频,给新员工培训用
Vector Animation
- - Create a vector-style animation that explains how attention works
- 用矢量动画讲清楚这个机制
Slides
- - Create an onboarding deck about MCP
- Make a teaching slide deck about prompt engineering
Image
- - Create a course cover image about MCP
- Make a social poster about prompt engineering
Web App
- - Build an interactive page that explains the MCP flow
- Create a small learning app about attention
Refinement examples
After the first result, the user may say things like:
- - make it shorter
- make it more polished
- switch it to Knowledge Video (Slides)
- switch it to vector animation
- keep it as an explainer video but make it feel more like a finished short video
- keep the same topic but change the tone
- make it more suitable for classroom use
Install options
OpenClaw / ClawHub
Install this skill natively in OpenClaw with:
CODEBLOCK0
Publish and registry workflows use ClawHub, while the runtime install path uses OpenClaw's native skills commands.
Skills CLI
Quick install while this repository contains one public skill:
CODEBLOCK1
Install this skill explicitly:
CODEBLOCK2
Install all skills in this repository explicitly:
CODEBLOCK3
You can preview discovery before installing:
CODEBLOCK4
Legacy Bun command
For teams that still use the old package name, the legacy Bun command is:
CODEBLOCK5
Prefer npx skills add for new documentation and examples.
Direct URL
Give the agent this file directly:
INLINECODE45
Local skill folder
For agents that support local skill folders, save this file as a local skill.
Example:
CODEBLOCK6
If the agent platform uses a different skills directory, save the same file there instead.
Where to read more
- - INLINECODE46
- INLINECODE47
- INLINECODE48
Security rules
- - the API key belongs to the user
- the agent should not ask the user to share the key with unrelated services
- the agent should not present itself as the KPainter account owner
- if the user resets the key, the agent should ask the user to reconnect it
Update guidance
If KPainter skills stop working or the user changes keys, the agent should:
- 1. reread this file or refresh the local skill
- ask the user to reconnect the latest API key
- confirm the available output types again
- ask what the user wants to create now
Success state
The setup is successful when the agent can:
- - guide the user through account and API key setup
- explain the available output types in simple language
- choose the right video style when the request is ambiguous, including
讲解视频 / 解说视频 / PPT视频 style requests, without treating bare video as an automatic Knowledge Video trigger - ask clarifying questions in normal language
- help the user create a first result and refine it
KPainter 技能
当智能体需要帮助用户将一个提示词转化为KPainter知识视频、幻灯片、图像或交互式应用时,尤其是在可控性、结构和知识传递比通用视频生成更重要的情况下,使用此技能。
官方网址
| 资源 | 网址 |
|---|
| 首页 | https://kpainter.ai/ |
| 技能文件 |
https://kpainter.ai/skill.md |
| 技能文档 | https://kpainter.ai/docs/skills |
| OpenAPI 文档 | https://kpainter.ai/docs/openapi |
| API 密钥 | https://kpainter.ai/api-key |
主要目标
- - 连接用户自己的KPainter账户和API密钥
- 帮助用户选择合适的输出类型
- 仅收集创建或优化结果所需的缺失信息
- 除非用户询问技术细节,否则使用日常语言交流
设置
如果KPainter尚未连接,请在尝试创建内容前引导用户完成设置。
- 1. 询问用户是否已有KPainter账户。
- 如果没有,告知用户在 https://kpainter.ai/ 注册或登录。
- 将用户引导至 https://kpainter.ai/api-key。
- 请用户激活、复制并将其API密钥连接到当前智能体。
- 只有在用户确认密钥已连接后,才表示设置完成。
不要要求用户与无关服务或在当前智能体连接流程之外分享其API密钥。
KPainter 可创建的内容
- - 知识视频
- 知识视频(幻灯片版)
- 矢量动画
- 幻灯片
- 图像
- 网页应用
智能体应如何选择输出类型
始终根据用户想要的结果进行路由。
- 1. 首先判断用户想要的是 知识视频 / 幻灯片 / 图像 / 网页应用。
- 如果用户说 知识视频、讲解视频 或 解说视频,先保持在视频类别内,不要直接跳到幻灯片或图像。
- 如果用户只说宽泛的 视频,用一个简短追问明确他们想要的是 知识视频、知识视频(幻灯片版) 还是 矢量动画。
- 然后在该类别内选择视频风格:
- 知识视频:适合精致、故事驱动、有旁白、适合社交媒体或接近成品短视频的内容
- 知识视频(幻灯片版):适合图解、静态、PPT风格、基于幻灯片、适合课堂、培训或分步骤讲解的内容
- 矢量动画:适合流程、机制、结构、工作流、原理、科学、数学或系统解释
- 5. 如果追问后视频类别内的需求仍然不明确,优先选择 知识视频(幻灯片版) 作为最安全的默认选项,并提供后续切换到更完整的 知识视频 风格。
- 当用户未指定时长时,知识视频 的默认目标长度约为 30秒。
- 除非用户明确询问,否则不要使用内部API类型名称。
触发指引
视频类别
将以下请求首先视为视频类别:
- - 制作一个知识视频
- 制作一个讲解视频
- 制作一个有旁白的讲解视频
- 帮我做一个讲解视频
- 帮我做一个解说视频
- 做一个知识视频把这件事讲清楚
知识视频
优先选择 知识视频 的请求:
- - 制作一个知识视频
- 让它更精致
- 让它更故事驱动
- 让它更适合社交媒体传播
- 让它感觉像成品短视频
- 制作一个有旁白的讲解视频
- 做一个更像成片的讲解视频
- 做一个更适合传播的解说视频
知识视频(幻灯片版)
优先选择 知识视频(幻灯片版) 的请求:
- - 图解版
- 静态
- PPT风格视频
- PPT视频
- 基于幻灯片的视频
- 类似幻灯片
- 分步骤讲解
- 课堂讲解
- 培训讲解
- 图解版
- PPT视频
- 课件视频
- 分步骤讲解
矢量动画
优先选择 矢量动画 的请求:
- - 矢量动画
- 流程动画
- 机制动画
- 架构动画
- 用动画展示这个流程或原理
- 用动画讲这个流程
- 用动画讲这个机制
幻灯片
- - 触发词如 幻灯片、PPT、演示文稿、讲稿、展示材料。
图像
- - 触发词如 图像、海报、封面、封面图、视觉摘要。
网页应用
- - 触发词如 应用、网页应用、交互页面、交互演示、学习应用。
积分不足时的回退指引
如果用户想要 知识视频 但积分不足,智能体应主动建议 知识视频(幻灯片版) 作为更便宜的回退方案。
推荐行为:
- 1. 告知用户当前的知识视频请求可能超出其可用积分。
- 提供 知识视频(幻灯片版) 作为低成本替代方案。
- 请求确认,而不是静默切换格式。
好的回退措辞:
- - 您当前的积分可能不足以制作完整的知识视频。我可以将其切换为知识视频(幻灯片版),这通常更便宜。您希望我这样做吗?
- 您当前的积分可能更适合制作知识视频(幻灯片版)。如果您愿意,我可以保持相同的主题,仅切换输出格式。
当用户接受时,保持相同的主题、受众和语言,仅切换输出格式,除非用户要求其他更改。
智能体应如何与用户交流
- - 每次只问一个后续问题。
- 保持问题简短具体。
- 尽可能使用日常语言而非技术术语。
- 仅帮助用户明确仍缺失的信息:主题、受众、语气、语言、时长、页数或宽高比。
- 如果结果不理想,帮助用户优化而非重新开始整个流程。
- 不要静默更改输出类型、受众、语言或格式。
- 如果当前平台上有文件或素材可用,将其视为用户的参考材料。
需收集的最少信息
仅询问仍缺失的信息:
- - 主题
- 受众
- 语言
- 时长或页数
- 语气或视觉方向
- 源文件或参考材料
如果用户已提供足够信息,先创建再优化。
多语言支持
KPainter 支持用户偏好的语言进行创建和优化工作流程。
用户可以用任何语言描述主题、受众、语气和约束条件。
智能体应保留用户要求的输出语言,除非用户要求切换。
如果预期的输出语言不明确,在创建前先问一个简短的后续问题。
多语言示例
仅为示例,非语言许可列表。
英语
Make a knowledge video that explains MCP clearly.
中文
帮我做一个讲清楚 MCP 的知识视频。
日语
MCP をわかりやすく説明する知識動画を作ってください。
阿拉伯语
أنشئ فيديو معرفيًا يشرح MCP بشكل واضح.
西班牙语
Crea un video de conocimiento que explique MCP con claridad.
韩语
MCP를 명확하게 설명하는 지식 영상을 만들어 주세요.
法语
Crée une vidéo de connaissance qui explique clairement le MCP.
用户请求示例
知识视频
- - 用知识视频解释什么是MCP
- 为新产品经理制作一个关于AI智能体的知识视频
- 制作一个约30秒的知识视频,清晰解释MCP
- 制作一个讲解视频,逐步介绍MCP
- 帮我做一个讲解视频,把MCP讲清楚
- 帮我做一个解说视频,适合发在社媒上
知识视频(幻灯片版)
- - 制作一个知识视频(幻灯片版),分步骤解释这个内容
- 做一个PPT视频,分步骤讲解MCP
- 做一个课件视频,给新员工培训用
矢量动画
- - 创建一个矢量风格的动画,解释注意力机制如何工作
- 用矢量动画讲清楚这个机制
幻灯片
- - 创建一个关于MCP的新员工入职演示文稿
- 制作一个关于提示工程的授课幻灯片
图像
- - 创建一个关于MCP的课程封面图
- 制作一个关于提示工程的社交媒体海报
网页应用
- - 构建一个解释MCP流程的交互页面
- 创建一个关于注意力机制的小型学习应用
优化示例
在首次结果之后,用户可能会说:
- - 让它更短
- 让它更精致
- 切换到知识视频(幻灯片版)
- 切换到矢量动画
- 保持为讲解视频,但让它更像成品短视频
- 保持相同主题但改变语气
- 让它更适合课堂使用
安装选项
OpenClaw / ClawHub
在OpenClaw中本地安装此技能:
bash
openclaw skills install kpainter
发布和注册工作流使用ClawHub,而运行时安装路径使用OpenClaw的原生技能命令。
Skills CLI
当此仓库包含