hypit
Clone any viral video with AI agents. Not just a script, the whole workflow: swap the face, the words, the B-roll, ship 100 variants in one command, and get your 100M views.
原始仓库简介 · 中文摘要待补充 · agentic-ai / ai / ai-agents / ai-video / compiler / dsl
README / 项目原文
Clone any viral video with AI agents
1 command, 100 variants, 100M views.
Demo • Quickstart • Develop • 简体中文
⭐ Help more people find Hypit and grow the community. Star this repo!
Hypit
Hypit gives AI agents (Claude Code, Codex...) a language and system to create video. Drop in a video, and your agent clones it as a complete workflow: footage, captions, B-roll and effects, all anchored to words instead of seconds.
To be clear: cloning a video is the fastest way in, not the only one. You can start from our templates, or just describe the video you want and your agent writes the workflow from scratch. Generation models are optional too: a workflow can compile captions, motion graphics and code-rendered visuals into a finished video without calling a generation model or incurring its service charges.
SVML source on the left, with the corresponding video rendered live on the right.
Install once
npx skills add hypit-ai/hypit -g
This installs the Skill. On first use, your agent checks for the Hypit executable and helps prepare it if needed. Your video project can live anywhere.
Hypit is free to use; your Coding Agent and model services have their own accounts and charges. HypiHub is our recommended hosted model service. You can also use your own API or local models; tell your agent the service name and API documentation so it can set up the appropriate connection.
Agent environments and entry partners · Model and deployment services
Examples
UGC
Generation source · Run and production notes
| Reference | |||
|---|---|---|---|
| Clones | |||
"GOAT DEBATE" — a 20-second football tier list that puts Ronaldo in D and Messi in S. Two Seedance 2 Mini 720p trolling A-rolls, a 2K goth girl portrait and ten 1K brain-rot B-rolls by GPT Image 2, WhisperX word alignment, a sound-synced ranking board, color-box karaoke captions, buttery-smooth animation and catchy background music, concurrently rendered in 64 headless Chromium processes. Three clones included: swap the narrator to banana cat, flip the rankings and Ronaldo becomes GOAT, or replace all players with tech founders — same viral structure, different viral video. Total cost: $1.15. |
|||
Podcast
Generation source · Run and production notes
| Reference | |||
|---|---|---|---|
| Clones | |||
"DAILY CREATINE" — an 18-second podcast clip where a muscle barbie forces creatine on a skinny college kid. Three confrontational A-rolls and an aspirational lifestyle B-roll by Seedance 2 Mini 720p, two 2K AI character portraits and three 1K montage panels by GPT Image 2, WhisperX word alignment, split-screen interview layout, speaker-aware karaoke captions, a product-handoff moment and background music, concurrently rendered in 64 headless Chromium processes. Three clones included: swap both hosts to Pepe and Doge arguing about doggy arms, replace the creatine with retinol where a pretty boy roasts a tomboy's pores, or ditch the physical product for a CheatGPT app where a himbo jock out-GPAs a PhD student — same podcast format, three ad verticals. Total cost: $1.07. |
|||
Street Interview
Generation source · Run and production notes
| Reference | |||
|---|---|---|---|
| Clones | |||
"NICE RIDE" — a 26-second street interview where a mob wife drops three rules for making her first million. Three A-rolls by Seedance 2 Mini 720p, one 2K AI character portrait by GPT Image 2, WhisperX word alignment, head-tracked speaker-colored captions driven by face bounding boxes from Google Video Intelligence and YOLOv8 AnimeFace, a sound-synced emoji reveal board with color flashes and reveal sound effects, plus background music, concurrently rendered in 64 headless Chromium processes. Three clones included: swap hosts to Wojak and Chad, translate everything to Spanish with exactly the same punchlines, or swap the Lambo for an F1 car where Ada explains to Leon how driving Uber led to an accidental Grand Prix win — same three-stage reveal, distinct street-interview remixes. Total cost: $1.09. |
|||
Use the Hypit skill
The /hypit skill is available to coding agents. Start a session in any empty or existing project
directory and ask it to create videos for you:
/hypit Clone this video: /path/to/video.mp4, and replace the ranking content with a comparison of Hypit (official website: hypit.ai) with other AI video products.
Or start without a reference video:
/hypit Make a ranking video that puts Hypit in S tier.
Your agent can check the environment, request the credentials the video needs, generate the material, and build the finished composition.
👉 Get 100 FREE AI Avatars with unique voices
Why Hypit
- Clone any video: drop in a video, get the whole workflow — footage, captions, B-roll, effects. Not a script breakdown.
- One workflow, 100 variants: reuse the composition and existing material; generate the parts that change.
- Pluggable components: swap the host without touching the captions. Use an installed package, adapt an example, or write your own.
- Open source: no Hypit seat pricing, per-render fees or added watermark. Model-service usage is billed by the service you choose.
What Hypit Can Build
Drop in a video and your agent clones the whole workflow — or describe what you want and it writes one from scratch. Either way you get an editable, re-runnable composition, not a one-off render.
- Paid social ads — clone a winning ad from the Meta Ad Library, swap in your product, ship 50 hook variants the same day. When it fatigues in two weeks, re-run with fresh openings; the body never changes.
- Viral clones — any TikTok, Reel, or Short becomes a template. Swap the host, the hook, the product, the language, the aspect ratio.
- TikTok Shop and affiliate videos — one format that converts, a new SKU every day. Swap the product, the price, the CTA; the structure that worked stays untouched.
- AI UGC and talking heads — narration, word-level captions, B-roll, comment stickers, beat-synced cuts, all wired automatically.
- Podcast and interview clips — split-screen layouts, speaker-aware captions, reaction overlays.
- Code-rendered videos — visuals driven entirely by front-end code, rendered locally without generation API calls.
- Localized versions — the same video in ten languages. Rewrite a line and the timing re-flows itself.
Stay Tuned with Us
Star us, and you will receive all release notifications from GitHub without any delay!
How to Contribute
Pull requests are welcome, and documentation, examples and translations count as much as code. Pick up an open issue or open one for what you want to work on, and we will help you land it. CONTRIBUTING.md has the setup, the checks CI runs and the pull request flow.
Creating a component is ordinary video-production work, and the component normally stays with the video project that owns it. When its owner wants to share it, the same package can be handed off as a versioned tarball or published under the owner's npm scope or private registry. Proposals for a component maintained in Hypit's official Distribution begin with an issue describing the shared product need; official inclusion is separate from normal community sharing.
The Development Guide covers the prerequisites, the daily commands and the repository layout.
| Bug reports | Open an issue |
| Feature requests | Open an issue |
| Questions | Discord or Telegram |
| Upgrading existing projects | 0.3 migration guide · 0.3.1 SDK migration guide |
Launch Partners
|
Watcha is an innovative ecosystem and community built for AI builders, connecting great products with high-quality users to turn every Demo into reality. Starting from AI product discovery and genuine feedback, Watcha provides product publishing, distribution, and reputation tracking to support builders from launch to early adoption. |
|
TokenDance offers builders a unified multi-model gateway with optimized inference costs. Through programs like the Billion-Token Subsidy and the Developer Program, TokenDance helps early-stage builders cut costs so they can focus on product validation and growth. |
|
AgentSky is a cloud platform for running AI agents such as Claude Code, Codex, OpenClaw, and Hermes. Each agent works on its own cloud computer, helping users tackle coding, research, and everyday tasks. Through CLI and MCP integrations, users can discover agents, delegate work, and retrieve results directly from their existing workflows. |
|
AutoClaw is Z.ai’s desktop AI assistant for work, deeply optimized for GLM models and integrated with specialized knowledge bases and workflows for legal, finance, design, and more. |
|
OpenAgents Workspace is a product that helps you manage AI agents, allowing you to invoke agents running on any device (whether on a mobile phone or computer) anytime, anywhere. Agents can collaborate within the platform to help you complete tasks and boost efficiency. It currently supports integration with over 20 types of agents, including Claude, Codex, OpenClaw, and Hermes. |
|
|
Monid is the OpenRouter for agent tools: one key, one wallet, hundreds of tools, billed per call. Hypit can route Seedance video, images, speech and word-level transcription to Monid. |
|
HiAPI is a unified API platform for AI image, video, audio, and language models. Connect GPT Image 2.5, Seedance 2.0, Veo 3.1, Kling, Wan, and other popular models through one platform. Pay as you go with no monthly fee, with free trial credits available for new users. |
|
Pollo API gives developers access to 300+ AI image and video models through a single API, with competitive rates. One API key connects you to multiple model providers and scales from 10 to 10,000+ requests. |
|
BeatAPI is the professional capability layer for any agent. We provide AI models, real-world data, task-ready capabilities, and professional workspaces that help agents complete real-world tasks. |
⭐ Star History
Contributors
License
Hypit is released under the Hypit Open Source License. The videos and other outputs you create belong to you; third-party models and services may have their own terms.
Partnership Community: LINUX DO
让 AI Agent 复刻任何爆款视频
一条命令,100 个变体,1 亿播放量。
⭐ 让更多人发现 Hypit,一起壮大社区。给仓库点个 Star!
Hypit
Hypit 为 AI Agent(Claude Code、Codex 等)提供了一套制作视频的语言和系统。给它一条视频,Agent 就能把它复刻成一份完整的 workflow:画面、字幕、B-roll、特效,全部锚定在词上,而不是秒上。
需要说明的是: 复刻视频是最快的入口,但不是唯一的入口。你可以从我们的模板开始,也可以描述你想要的视频,让 Agent 从零写出一份 workflow。生成模型同样不是必需的:字幕、动效和代码渲染的画面,不调用生成模型也能编译成一条成片,因此可以不产生模型服务费用。
左侧是 SVML 源文件,右侧实时呈现对应视频。
只安装一次
npx skills add hypit-ai/hypit -g
这条命令安装 Skill。首次使用时,Agent 会检查 Hypit 可执行程序,并按需协助安装。 视频项目可以放在任意位置。
Hypit 本身免费使用;Coding Agent 和模型服务各有自己的账号与费用。 HypiHub 是我们推荐的托管模型服务,也可以使用你自己的 API 或本地模型。 把服务名称和 API 文档告诉 Agent,它会据此配置合适的连接。
示例
UGC
Generation source · Run and production notes
| 参考视频 | |||
|---|---|---|---|
| 复刻视频 | |||
“GOAT DEBATE”——一条 20 秒的足球 Tier List,把 Ronaldo 排进 D 级,把 Messi 排进 S 级。A-roll 是两段 Seedance 2 Mini 生成的 720p 吐槽片段,B-roll 是 GPT Image 2 生成的一张 2K 哥特女孩肖像和十张 1K 脑腐图;WhisperX 负责逐词对齐,排行榜跟着音效逐条落位,再配上彩色词盒卡拉 OK 字幕、丝滑的动画和抓耳的背景音乐,最后由 64 个无头 Chromium 进程并发渲染。 上面给出了三个复刻版本:把解说员换成香蕉猫;翻转排名,让 Ronaldo 成为 GOAT;把所有球员换成科技公司创始人。同一套爆款结构,可以产出完全不同的爆款视频。 总成本:$1.15。 |
|||
播客
Generation source · Run and production notes
| 参考视频 | |||
|---|---|---|---|
| 复刻视频 | |||
“DAILY CREATINE”——一条 18 秒的播客片段,肌肉芭比向一个瘦弱的大学生强推肌酸。A-roll 是 Seedance 2 Mini 生成的三段 720p 对峙镜头,外加一段展示理想生活方式的 B-roll;图像是 GPT Image 2 生成的两张 2K AI 角色肖像和三张 1K 蒙太奇画面。WhisperX 负责逐词对齐,再配上分屏访谈版式、区分说话人的卡拉 OK 字幕、产品递出的瞬间和背景音乐,最后由 64 个无头 Chromium 进程并发渲染。 上面给出了三个复刻版本:把两位主播换成 Pepe 和 Doge,争论狗狗手臂;把肌酸换成视黄醇,让漂亮男孩吐槽假小子的毛孔;把实体产品换成 CheatGPT 应用,让头脑简单的体育生绩点碾压博士生。同一套播客形式,三个广告垂类。 总成本:$1.07。 |
|||
街头采访
Generation source · Run and production notes
| 参考视频 | |||
|---|---|---|---|
| 复刻视频 | |||
“NICE RIDE”——一条 26 秒的街头采访,黑帮太太分享赚到第一个一百万的三条规则。三段由 Seedance 2 Mini 生成的 720p A-roll,一张由 GPT Image 2 生成的 2K AI 角色肖像;WhisperX 负责逐词对齐,再利用 Google Video Intelligence 和 YOLOv8 AnimeFace 提供的人脸边界框,驱动跟随人脸的说话人专属彩色字幕;配合跟音效同步的表情符号揭示板、彩色闪烁、揭示音效和背景音乐,最后由 64 个无头 Chromium 进程并发渲染。 上面给出了三个复刻版本:把主播换成 Wojak 和 Chad;把所有内容翻译成西班牙语,同时保持完全相同的 punchline;或者把兰博基尼换成 F1 赛车,让 Ada 向 Leon 解释自己如何因为开 Uber 意外赢得大奖赛。同样的三阶段揭示结构,三种不同的街头采访混剪。 总成本:$1.09。 |
|||
使用 Hypit skill
编程 Agent 可以直接使用 /hypit skill。在任意空目录或现有项目目录中开启会话,让 Agent 为你创建视频:
/hypit 复刻这个视频:/path/to/video.mp4,把排行榜内容换成 Hypit(官网:hypit.ai)与其他 AI 视频产品的对比。
也可以不提供参考视频,直接从描述开始:
/hypit 做一个 ranking 视频,把 Hypit 排到 S 级。
Agent 会检查环境,索要这条视频需要的凭据,生成素材并构建成片。
👉 免费获得 100 个拥有独特音色的 AI 人物形象
为什么选择 Hypit
- 真的能 Clone: 丢一条视频进来,拿到整份 workflow —— 画面、字幕、B-roll、特效。不是拆解脚本。
- 一份 workflow,100 个变体: 复用编排和已有素材,需要变化的部分再生成。
- 可插拔组件: 换主播不动字幕。使用已安装的包、改造一个示例,或自己编写。
- 开源: Hypit 不按人头或渲染次数收费,也不添加水印。模型服务费用由你选择的服务方收取。
Hypit 能构建什么
丢一条视频进来,Agent 把整套 workflow 克隆下来;或者直接描述你想要的视频,让它从零写一份。拿到的都是可编辑、可重跑的工作流,而不是一次性的成片。
- 信息流广告 —— 从 Meta 广告库拉一条正在跑的广告,克隆成 workflow,换上你的产品,当天发出 50 个 Hook 变体。两周后创意衰退,换一批开头重跑,主体一个字不动。
- 复刻爆款 —— 任何 TikTok、Reels、Shorts 都能变成模板:换主播、换 Hook、换产品、换语言、换画幅。
- 带货视频 —— 一个跑通的格式,每天换一个 SKU。换产品、换价格、换 CTA,转化率跑出来的结构原封不动。
- AI UGC 与口播 —— 配音、词级字幕、B-roll、评论区贴纸、卡点剪辑,全部自动挂好。
- 播客与采访切片 —— 分屏布局、区分说话人的字幕、反应特效。
- 代码渲染的视频 —— 画面完全由前端代码驱动,本地渲染,无需调用生成模型 API。
- 多语言版本 —— 同一条视频十种语言。改一句台词,时间轴自己重排。
关注我们
给我们点个 Star,即可第一时间收到 GitHub 的所有版本发布通知!
贡献指南
欢迎提交 Pull Request,文档、示例和翻译与代码同样重要。可以认领一个已有的 issue,也可以为你想做的事新开一个,我们会协助你把它合并进来。环境准备、CI 会跑的检查和 Pull Request 流程见 CONTRIBUTING.zh-CN.md。
创建组件是正常的视频制作工作,组件通常留在拥有它的视频项目中。所有者希望分享时,可以把同一份包作为版本化 tarball 直接交付,也可以发布到自己的 npm scope 或私有 registry。希望某个组件由 Hypit 官方 Distribution 维护时,先通过 issue 说明它解决的共同产品需求;官方收录与普通的社区分享是两件事。
开发指南说明了前置条件、日常命令和仓库结构。
| 缺陷报告 | 提交 issue |
| 功能请求 | 提交 issue |
| 问题咨询 | Discord 或 Telegram |
| 旧项目升级 | 0.3 迁移指南 · 0.3.1 SDK 迁移指南 |
扫码加入微信群:

Launch 伙伴
|
观猹是一个专为 AI 开发者打造的创新产品生态与社区,致力于连接优秀产品与高质量用户,助力每一个想法从 Demo 走向真实落地。从产品发现与真实反馈起步,观猹提供发布、分发与声誉追踪等支持,全方位陪伴开发者完成从产品上线到首批用户获取的全过程。 |
|
TokenDance 为开发者提供优化了推理成本的统一多模型网关。通过百亿 Token 补贴和开发者计划,TokenDance 帮助早期开发者大幅降低成本,更专注于产品验证与业务增长。 |
|
AgentSky 是一个用于运行 Claude Code、Codex、OpenClaw 和 Hermes 等 AI 智能体的云平台。每个智能体都在独立的云端计算机上运行,帮助用户完成编程、研究和日常任务。通过 CLI 和 MCP 集成,用户可以直接在现有工作流程中发现智能体、委派任务并获取结果。 |
|
AutoClaw 是智谱推出的桌面端 AI 工作助手,深度适配 GLM 模型,融合法律、金融、设计等领域的专属知识库与工作流。 |
|
OpenAgents Workspace 是一个让你管理智能体的产品,它能让你随时随地调用在任何设备上运行的 Agent(手机 + 电脑端均可)。Agents 可以在其中协同合作,帮你完成任务、提升效率。现支持 Claude、Codex、Openclaw、Hermes 等 20 余种 agent 接入。 |
|
|
Monid 是 Agent 工具的 OpenRouter:一个 key、一个钱包、数百种工具,按调用计费。Hypit 可以把 Seedance 的视频、图像、语音和词级转写路由到 Monid。 |
|
HiAPI 是一个统一的 AI 图片、视频、音频和语言模型 API 平台,支持 GPT Image 2.5、Seedance 2.0、Veo 3.1、可灵、万相等主流模型。HiAPI 提供在线 Playground、Remote MCP 和 Agent Skills,可接入 Claude Code、Cursor 等 AI 编程工具。平台按量付费,无月费,新用户注册可获得免费体验积分。 |
|
Pollo API 通过单个 API 为开发者提供 300 多个 AI 图片与视频模型,价格具备竞争力。一个 API key 即可接入多家模型服务商,请求规模从 10 次到 10000 次以上均可承载。 |
|
BeatAPI 是面向各类 Agent 的专业能力层,提供 AI 模型、真实世界数据、可直接调用的任务能力和专业工作空间,帮助 Agent 完成真实世界的任务。 |
⭐ Star History
贡献者
许可证
Hypit 采用 Hypit 开源许可证。你创作的视频和其他产出归你所有;第三方模型与服务可能另有条款。




