Video Editor Skill

Automated video editing tool that submits Alibaba Cloud editing tasks based on provided materials and editing requirements, without requiring ffmpeg installation, waits for task completion, and outputs the final video URL.

Core Design Philosophy

This skill adopts a separation of concerns design:

1. 1. references/ - LLM knowledge base containing best practice documentation for various scenarios
2. scripts/ - Pure execution tools responsible only for submitting tasks and polling status

The LLM should refer to documents in references/ to generate Timeline JSON in Alibaba Cloud ICE format, then use scripts to submit tasks.

Prerequisites

Pre-check: Install Python Dependencies
CODEBLOCK0

Pre-check: Alibaba Cloud Credentials Required

Scripts automatically obtain credentials via the Alibaba Cloud default credential chain, supporting the following methods (in priority order):

1. 1. Environment variable credentials
2. Configuration file: INLINECODE1
3. ECS RAM Role (when running on ECS)

It is recommended to use the aliyun configure command to set up credentials:

> aliyun configure
> 

Or refer to the Alibaba Cloud Credential Configuration Documentation to configure the default credential chain.

OSS Bucket Configuration

OSS upload functionality requires Bucket information to be configured via environment variables:

> export OSS_BUCKET=your_bucket_name
> export OSS_ENDPOINT=oss-cn-shanghai.aliyuncs.com
> 

If OSS_BUCKET is not configured, list the buckets under the customer's current account and let the customer choose one as the output bucket for the final video.

OSS operations reuse the Alibaba Cloud default credential chain; no separate OSS credential configuration is needed.

User-Agent Configuration

All Alibaba Cloud service calls must set User-Agent to AlibabaCloud-Agent-Skills.
The script scripts/video_editor.py has already automatically configured this User-Agent.

Workflow

Step 1: Understand User Requirements

Analyze the type of video the user wants to create:

• - Slideshow video (image carousel)
• Multi-track audio mixing (voiceover + music)
• Multi-clip stitching
• Add subtitles/titles
• Effects and transitions
• Picture-in-picture/split-screen effects

Step 2: Reference Best Practices

Consult the corresponding documents in references/:

Step 3: Prepare Material URLs

• - If it is a local file, you need to call the oss-upload skill to upload it and obtain the OSS URL, which can be directly spliced into the timeline
• If you already have a URL, you can directly splice it into the timeline

Step 4: Generate Timeline JSON

Generate a Timeline in Alibaba Cloud ICE format according to the reference documents:

CODEBLOCK3

Step 5: Submit Editing Task

Use the script to submit the task (based on Alibaba Cloud Common SDK):

CODEBLOCK4

Parameter Description:

OutputMediaConfig example:

{
  "MediaURL": "https://{your-bucket}.oss-cn-shanghai.aliyuncs.com/{your-target-video-path}",
  "Width": 1080,  
  "Height": 1920
}

After the task is submitted, a JobId will be returned.

Step 6: Poll Task Status

Use the script to query/wait for task completion:
CODEBLOCK6

When the task status is Success, call GetMediaInfo based on the returned MediaId to obtain the OSS URL with authentication, and return it.

Timeline Example

Simplest Slideshow

CODEBLOCK7

LLM Prompt Suggestions

When generating a Timeline, think like this:

1. 1. What type of video does the user need? → Find the corresponding reference document
2. Which tracks are needed? (video track, audio track, subtitle track)
3. What clips are in each track?
4. Do you need to set In/Out/TimelineIn/TimelineOut (simple stitching does not require setting)? If setting is needed, what are In/Out/TimelineIn/TimelineOut respectively?
5. Are effects, transitions, volume adjustments, etc. needed?
6. Generate the complete JSON

Related Files

CODEBLOCK8