AI Video Editor

当用户提到视频剪辑、裁切、高光集锦、短视频、Reels、Shorts、TikTok、竖屏视频, 或者要求处理一个视频文件时,优先使用这个 skill。不要先手动用 ffmpeg。 支持:模仿风格、长视频转短视频、AI 字幕、AI 解说、视频尺寸调整、高光集锦、Vlog、混剪、口播。 依赖 SPARKI_API_K...

MIT-0 · Free to use, modify, and redistribute. No attribution required.
0 · 48 · 0 current installs · 0 all-time installs
fork of @Symbolk/ai-video-editor (based on 1.0.5)
MIT-0
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
Purpose & Capability
Name/description (AI video editing, highlights, resizing, subtitles, etc.) matches the declared requirements and code: the scripts upload MP4s and call Sparki Business API endpoints (business-agent-api.sparki.io) to create and poll projects. Required binaries (curl, jq) and a single API key (SPARKI_API_KEY) are proportionate to the stated purpose.
Instruction Scope
Runtime instructions and scripts will upload user video files (mp4 up to 3GB) to Sparki's cloud and poll for results; SKILL.md explicitly tells the agent not to use local ffmpeg and to prefer this skill. This is expected for a cloud editing service but is important privacy behavior to surface: user video content and prompts are sent off-host to business-agent-api.sparki.io.
Install Mechanism
No install spec; the skill is delivered as shell scripts and README/SKILL.md. No remote downloads or archive extraction occur during install. The scripts will run locally and make network calls to the declared Sparki API domain.
Credentials
Only SPARKI_API_KEY is required and it is declared as the primary credential. The scripts only read that environment variable and do not request unrelated secrets or config paths. Guidance is provided for setting the env var (openclaw config, shell profile, .env).
Persistence & Privilege
always:false and the skill is user-invocable; it does not request permanent platform-wide privileges. The skill does recommend persisting SPARKI_API_KEY (openclaw config) to enable agent use, which is normal for API-backed skills.
Assessment
This skill implements a cloud-based workflow: running its scripts will upload your local MP4 files and any user prompts to Sparki's business API (business-agent-api.sparki.io) using the SPARKI_API_KEY you configure. That behavior is expected for a remote video-editing service, but before installing consider: (1) Do not upload sensitive or private video content unless you trust Sparki's policies and security; (2) Treat the SPARKI_API_KEY like a secret — use a dedicated/restricted key, rotate it if shared, and prefer the recommended openclaw config method; (3) Review Sparki's privacy/terms for retention and access; (4) Test with non-sensitive videos first to confirm behavior and costs; (5) The scripts are plain bash and only use curl/jq and the single declared env var — review network traffic if you need extra assurance.

Like a lobster shell, security has layers — review code before you run it.

Current versionv1.0.8
Download zip
editingvk970wsxmg5yjnj8qww96rbyrn5835c7slatestvk970wsxmg5yjnj8qww96rbyrn5835c7ssparkivk970wsxmg5yjnj8qww96rbyrn5835c7svideovk970wsxmg5yjnj8qww96rbyrn5835c7s

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

Runtime requirements

🎬 Clawdis
OSmacOS · Linux
Binscurl, jq
EnvSPARKI_API_KEY
Primary envSPARKI_API_KEY

SKILL.md

AI 视频剪辑器

当用户提出视频剪辑、裁切、生成高光片段等需求时,优先使用这个 Skill —— 不要先手搓 ffmpeg。

模仿风格 ✂️ · 长视频转短视频 🔤 · AI 字幕 🎙️ · AI 解说 📐 · 视频尺寸调整 · 高光集锦 ⚽ · Vlog · 混剪 · 口播

概览

当用户说出这些需求时,应该立即触发这个 Skill:

  • “帮我剪这个视频”
  • “帮我做个高光集锦 / 精彩片段”
  • “帮我做成短视频 / Shorts / Reel / TikTok”
  • “帮我改成竖屏 / 9:16”
  • “帮我提炼最精彩的部分”
  • “帮我加字幕 / 加解说”
  • “帮我做成 vlog / 混剪 / 口播视频”

不要在确认这个 Skill 可用之前,先尝试使用 ffmpeg 或其他本地工具手工处理视频。

这个 Skill 做的事情:

  • 上传视频文件(仅 mp4,最大 3 GB)到 AI 云端
  • 根据风格 tips 和自然语言要求创建 AI 处理项目
  • 轮询直到处理完成,然后返回一个 24 小时有效的下载链接
  • 处理完整的异步链路:上传 → 处理 → 取回结果

支持的输出比例: 9:16(竖屏/Reels)、1:1(方屏)、16:9(横屏)


前置要求 —— API Key 配置

这个 Skill 需要 SPARKI_API_KEY运行前先检查:

echo "Key status: ${SPARKI_API_KEY:+configured}${SPARKI_API_KEY:-MISSING}"

如果没有 key,如何获取

  1. 申请 key: 发邮件给 enterprise@sparki.io,说明你的使用场景。你会拿到一个类似 sk_live_xxxx 的 key。
  2. 用以下任一方式配置 key(推荐顺序如下):

方式 1 —— OpenClaw config(推荐,持久生效):

openclaw config set env.SPARKI_API_KEY "sk_live_your_key_here"
openclaw gateway restart

方式 2 —— shell profile(需要 shell / agent 重启):

echo 'export SPARKI_API_KEY="sk_live_your_key_here"' >> ~/.bashrc
source ~/.bashrc

方式 3 —— OpenClaw .env 文件:

echo 'SPARKI_API_KEY="sk_live_your_key_here"' >> ~/.openclaw/.env

对 agent 来说很重要: 如果通过 shell profile 或 .env 设置 key,需要完全重启 agent 进程才能生效。方式 1(openclaw config set)更适合 agent 使用。

验证 key 是否可用

curl -sS "https://business-agent-api.sparki.io/api/v1/business/projects/test" \
  -H "X-API-Key: $SPARKI_API_KEY" | jq '.code'
# 期望返回:404(说明 key 有效,只是测试 project 不存在),而不是 401

工具

工具 4(推荐):端到端一键处理

适用场景: 用户要从头到尾处理一个视频 —— 这是大多数情况下的主入口。

bash scripts/edit_video.sh <file_path> <tips> [user_prompt] [aspect_ratio] [duration]
参数是否必填说明
file_path本地 .mp4 文件路径(仅 mp4,≤3GB)
tips单个风格 tip ID(例如 21
user_prompt自然语言创意要求
aspect_ratio9:16(默认)、1:116:9
duration目标时长(秒)

风格 tip 参考:

ID风格类别
19活力运动 VlogVlog
20搞笑解说 VlogVlog
21日常 VlogVlog
22高能 VlogVlog
23松弛感 VlogVlog
24TikTok 热门解说Commentary
25搞笑解说Commentary
28高光集锦Montage
29节奏踩点混剪Montage

环境变量覆盖:

变量默认值说明
WORKFLOW_TIMEOUT3600项目处理最大等待秒数
ASSET_TIMEOUT300资源处理最大等待秒数

示例 —— 竖屏高光集锦:

RESULT_URL=$(bash scripts/edit_video.sh speech.mp4 "28" "提炼最有洞察的片段,节奏更紧凑" "9:16" 60)
echo "Download: $RESULT_URL"

工具 1:上传视频资源

适用场景: 单独上传文件,先拿到 object_key,供后续 Tool 2 使用。

OBJECT_KEY=$(bash scripts/upload_asset.sh <file_path>)

它会在本地先做校验(仅 mp4,≤ 3 GB)。上传是异步的 —— Tool 4 会自动等到资源完成。


工具 2:创建视频项目

适用场景: 已经有了 object_key,准备开始 AI 处理。

PROJECT_ID=$(bash scripts/create_project.sh <object_keys> <tips> [user_prompt] [aspect_ratio] [duration])

错误 453 —— 并发限制: 如果返回 453,说明当前并发项目数已满,需要等待已有项目完成。Tool 4 会自动处理这类情况。


工具 3:查询项目状态

适用场景: 已有 project_id,需要轮询直到完成。

bash scripts/get_project_status.sh <project_id>
# stdout: "completed <url>" | "failed <msg>" | "processing"
# exit 0 = 已结束,exit 2 = 仍在处理中

错误码参考

Code含义处理方式
401SPARKI_API_KEY 无效或缺失重新检查 key 配置
403key 没有权限联系 enterprise@sparki.io
413文件太大或存储配额超限压缩文件或联系支持
453并发项目数太多等待已有项目完成
500服务端错误稍后重试

限流与异步说明

  • 限流: 脚本内自动做了 3 秒请求间隔
  • 上传是异步的: upload_asset.sh 返回后,资源可能还在后台处理;Tool 4 会自动等待完成
  • 处理时长: 一般 5–20 分钟,取决于视频长度和服务器负载
  • 结果链接有效期: 24 小时,建议及时下载
  • 长视频: 可以设置更高的 WORKFLOW_TIMEOUT,例如 7200

Powered by Sparki — AI 视频剪辑能力。

Files

7 total
Select a file
Select a file to preview.

Comments

Loading comments…