clipspeed-mcp
Enables turning any video into scored, captioned, ready-to-post vertical clips directly from your MCP client, with capabilities to download finished MP4s, schedule posts, and discover trending content.
README
clipspeed-mcp
Turn any video into scored, captioned, ready-to-post vertical clips — right inside Claude, Cursor, Cline, or Windsurf. clipspeed-mcp is a stdio MCP server that wraps the ClipSpeed public API with your personal API key, and (unlike a hosted MCP) can download the finished .mp4 clips straight to your ~/Downloads folder.
Requires a ClipSpeed paid plan or the $1 trial. Get your key at https://www.clipspeed.ai/dashboard/api-key/ — keys look like csai_live_....
Always pin
clipspeed-mcp@1.0.1. Pinning keeps a bad publish from silently upgrading your setup.
Install
Claude Desktop (one command)
npx clipspeed-mcp@1.0.1 install
It prompts for your API key and writes the clipspeed server into your Claude Desktop config for you. You can also pass the key directly:
npx clipspeed-mcp@1.0.1 install --key csai_live_your_key
Then quit and reopen Claude Desktop.
Claude Code
claude mcp add clipspeed -- npx -y clipspeed-mcp@1.0.1
export CLIPSPEED_API_KEY=csai_live_your_key
Cursor / Cline / Windsurf
Add this to your MCP servers config (~/.cursor/mcp.json, Cline's MCP settings, or Windsurf's mcp_config.json):
{
"mcpServers": {
"clipspeed": {
"command": "npx",
"args": ["-y", "clipspeed-mcp@1.0.1"],
"env": { "CLIPSPEED_API_KEY": "csai_live_your_key" }
}
}
}
Hosted (no install)
Prefer a remote server with no local process? ClipSpeed also exposes a hosted MCP:
claude mcp add --transport http clipspeed https://api.clipspeed.ai/mcp \
--header "Authorization: Bearer csai_live_your_key"
The hosted server can do everything except write files to your disk — use this local package if you want download_clip.
Environment
| Var | Required | Default | Notes |
|---|---|---|---|
CLIPSPEED_API_KEY |
yes (to serve) | — | Your personal csai_live_ key. |
CLIPSPEED_BASE_URL |
no | https://api.clipspeed.ai |
Override for testing. |
Node >= 18 (uses global fetch).
Tools
| Tool | Does |
|---|---|
clip_video |
Video URL → a clip job (job_id). Options: aspect_ratio, caption_style, clip_count, platform. |
get_clip_status |
Poll a job_id → status + finished clips (title, viral_score, duration, video_url). |
list_clips |
List your recent clip jobs to find a past job_id. |
download_clip |
Save the .mp4 files to ~/Downloads — all clips, or one by clip_index. Skips files > 500MB. |
get_usage |
Credits remaining / total / used this month, plan, trial status. |
discover_trending |
Find the fastest-growing recent video in a niche to clip. ClipSpeed's differentiator. |
schedule_post |
Schedule a finished clip to tiktok / youtube / instagram / x / linkedin. |
list_scheduled_posts |
Review scheduled / posted / failed posts. |
cancel_scheduled_post |
Cancel a post that's still scheduled. |
restyle_captions |
Re-caption a project in a new style without re-clipping. |
Caption styles: karaoke (default), hormozi, beasty, fire, youshaei, cinematic, bold.
Aspect ratios: 9:16 (default), 1:1, 16:9.
To let ClipSpeed publish to your socials, connect each platform in the ClipSpeed app first (Scheduler). schedule_post posts under those connections.
The demo prompt
Clip this video into 9:16 shorts with Hormozi captions, then download the top 3 to my Downloads folder: https://youtu.be/…
Claude runs clip_video → polls get_clip_status → download_clip → the .mp4s land in ~/Downloads.
A Claude Skill (SKILL.md) ships with this package — it teaches the outcome-first flow (find → clip → check → download → schedule).
Security
- Your API key is never logged. It only ever lands in your MCP client's config file (or your shell env).
- Never commit your key or share it. Rotate it any time at https://www.clipspeed.ai/dashboard/api-key/.
- Keys are tied to one account, rate-limited per plan, and hashed at rest on ClipSpeed's servers.
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。