HappyHorse Video Generation and Editting

Video Generation and Editing with HappyHorse models. Supports text2video, image2video (first-frame based), reference2video (multi-image character fusion), an...

Security Scan

The skill's code, docs, and required credential match its stated purpose (HappyHorse / DashScope video generation); requested API key and python are appropriate and proportional.

Purpose & Capability

Name/description, referenced endpoints (dashscope.aliyuncs.com), and required environment variable (DASHSCOPE_API_KEY) align with the HappyHorse / DashScope AIGC video service; python3 requirement is expected for the provided script.

Instruction Scope

Runtime instructions tell the agent to run the included Python script which performs HTTP requests to dashscope.aliyuncs.com. The script will accept URLs or local file paths and will base64-embed local files (data: URIs) into the request body — this is necessary to upload local media but means arbitrary local files can be read and transmitted if the user or agent supplies those paths. The SKILL.md does not attempt to read unrelated system config or credentials.

Install Mechanism

There is no install spec (instruction-only with an included script). Nothing is downloaded from unknown URLs and no installers run; the only runtime dependency is python3 (the script also uses the requests library, which is not installed by the skill).

Credentials

Only one environment variable is required: DASHSCOPE_API_KEY (declared as primaryEnv) which is appropriate for authenticating to the DashScope API. No unrelated credentials or config paths are requested.

Persistence & Privilege

The skill is not forced-always-enabled and does not request elevated or cross-skill config changes. Note: autonomous invocation is allowed by default (platform default) — combined with the ability to upload local files, this increases the practical blast radius if the agent is granted broad autonomy.

Assessment

This skill appears to do what it claims: it calls DashScope (HappyHorse) APIs and requires a DASHSCOPE_API_KEY. Before installing: 1) Only provide a DashScope API key you trust and that has limited scope; avoid using high-privilege or long-lived keys. 2) Be careful with local-file arguments: the included script will base64-embed and send local files to the external service, so do not point it at sensitive files (SSH keys, credentials, private documents). 3) Ensure the runtime environment has python3 and a safe requests installation (the script assumes the requests package). 4) If you enable autonomous agent invocation, consider limiting autonomy or auditing requests and outputs — autonomous agents could be instructed to upload arbitrary local files. 5) If you need higher assurance, validate the API endpoints and behavior by running the script in an isolated environment (container/VM) and inspect network traffic or server-side logs.

HappyHorse Video Models

HappyHorse Video Models, created by Alibaba Group, are video generation and editing models. This skill integrates with HappyHorse Model APIs on ModelStudio (Bailian-Alibaba Model Service Platform).

Model Overview

Model	Capabilities	Resolution	Duration
happyhorse-1.0-t2v	Text to Video	720P, 1080P	3-15s
happyhorse-1.0-i2v	First-frame Image to Video	720P, 1080P	3-15s
happyhorse-1.0-r2v	Reference images to video, Multi-character fusion	720P, 1080P	3-15s
happyhorse-1.0-video-edit	Video editing, Style transfer, Reference-based edit	720P, 1080P	3-15s (output)

Text to Video Generation (happyhorse-1.0-t2v)

Generate videos from text prompts.

text2video task-submit

python3 {baseDir}/scripts/happyhorse-magic.py text2video-gen --prompt "一座由硬纸板和瓶盖搭建的微型城市，在夜晚焕发出生机" --resolution "720P" --ratio "16:9" --duration 5
python3 {baseDir}/scripts/happyhorse-magic.py text2video-gen --prompt "一只可爱的小猫将军站在悬崖上，远处是壮丽的雪山" --resolution "1080P" --ratio "16:9" --duration 10
python3 {baseDir}/scripts/happyhorse-magic.py text2video-gen --prompt "海浪拍打岩石的慢镜头" --resolution "720P" --ratio "9:16" --duration 8 --no-watermark

Options

• --prompt: Text prompt for video generation (required, max 5000 non-Chinese chars or 2500 Chinese chars)
• --resolution: Video resolution - 720P or 1080P (default: 1080P)
• --ratio: Aspect ratio - 16:9, 9:16, 1:1, 4:3, 3:4 (default: 16:9)
• --duration: Video duration in seconds, 3-15 (default: 5)
• --no-watermark: Disable "Happy Horse" watermark (default: enabled)
• --seed: Random seed for reproducibility [0, 2147483647]

text2video tasks-get (round-robin)

python3 {baseDir}/scripts/happyhorse-magic.py text2video-get --task-id "<TASK_ID_FROM_VIDEO_GEN>"

Image to Video Generation (happyhorse-1.0-i2v)

Generate video from a first-frame image. Output aspect ratio automatically follows the input image.

First-frame to Video

python3 {baseDir}/scripts/happyhorse-magic.py image2video-gen --prompt "一只猫在草地上奔跑" --first-frame "https://example.com/cat.png" --resolution "720P" --duration 5
python3 {baseDir}/scripts/happyhorse-magic.py image2video-gen --first-frame "https://example.com/landscape.jpg" --resolution "1080P" --duration 10
python3 {baseDir}/scripts/happyhorse-magic.py image2video-gen --prompt "花朵缓缓绽放" --first-frame "https://example.com/flower.png" --duration 8 --no-watermark

Options

• --prompt: Text prompt for video generation (optional but recommended)
• --first-frame: First frame image URL (required). Format: JPEG/JPG/PNG/WEBP, min 300px per side, max 10MB.
• --resolution: Video resolution - 720P or 1080P (default: 1080P). Output ratio follows input image.
• --duration: Video duration in seconds, 3-15 (default: 5)
• --no-watermark: Disable "Happy Horse" watermark (default: enabled)
• --seed: Random seed for reproducibility [0, 2147483647]

Note: There is no --ratio parameter. The output video aspect ratio automatically follows the input first-frame image.

image2video tasks-get (round-robin)

python3 {baseDir}/scripts/happyhorse-magic.py image2video-get --task-id "<TASK_ID_FROM_VIDEO_GEN>"

Reference to Video Generation (happyhorse-1.0-r2v)

Generate video from reference images as character/object sources. Use "character1", "character2", etc. in the prompt to reference images by their array order.

Multi-character Reference

python3 {baseDir}/scripts/happyhorse-magic.py reference2video-gen --prompt "身着红色旗袍的女性character1，镜头以侧面中景勾勒旗袍修身剪裁" --reference-images "https://example.com/girl.jpg" --resolution "720P" --ratio "16:9" --duration 5
python3 {baseDir}/scripts/happyhorse-magic.py reference2video-gen --prompt "身着红色旗袍的女性character1，轻抬玉手展开折扇character2时流苏耳坠character3随头部转动轻盈摆动" --reference-images "https://example.com/girl.jpg" "https://example.com/fan.jpg" "https://example.com/earrings.jpg" --resolution "720P" --ratio "16:9" --duration 5
python3 {baseDir}/scripts/happyhorse-magic.py reference2video-gen --prompt "character1在海边散步，阳光洒在沙滩上" --reference-images "https://example.com/person.png" --resolution "1080P" --ratio "16:9" --duration 10

Options

• --prompt: Text prompt with character references (required, max 5000 non-Chinese chars). Use "character1/character2/..." to reference images in array order.
• --reference-images: Reference image URLs (required, 1-9 images). The 1st image = character1, 2nd = character2, etc.
• --resolution: Video resolution - 720P or 1080P (default: 1080P)
• --ratio: Aspect ratio - 16:9, 9:16, 1:1, 4:3, 3:4 (default: 16:9)
• --duration: Video duration in seconds, 3-15 (default: 5)
• --no-watermark: Disable "Happy Horse" watermark (default: enabled)
• --seed: Random seed for reproducibility [0, 2147483647]

Note: Reference images should have short side >= 400px. Avoid low-resolution, blurry, or heavily compressed images.

reference2video tasks-get (round-robin)

python3 {baseDir}/scripts/happyhorse-magic.py reference2video-get --task-id "<TASK_ID_FROM_VIDEO_GEN>"

Video Editing (happyhorse-1.0-video-edit)

Edit existing videos with text instructions and optional reference images. Supports style transfer, local replacement, and other editing tasks.

Instruction-based Editing

python3 {baseDir}/scripts/happyhorse-magic.py video-edit --prompt "将场景变成夜晚，添加霓虹灯效果" --video "https://example.com/video.mp4" --resolution "720P"
python3 {baseDir}/scripts/happyhorse-magic.py video-edit --prompt "为人物换上酷闪的衣服" --video "https://example.com/original.mp4"

Reference Image-based Editing

python3 {baseDir}/scripts/happyhorse-magic.py video-edit --prompt "让视频中的角色穿上图片中的条纹毛衣" --video "https://example.com/video.mp4" --reference-images "https://example.com/clothes.webp"
python3 {baseDir}/scripts/happyhorse-magic.py video-edit --prompt "将背景替换为参考图中的场景" --video "https://example.com/video.mp4" --reference-images "https://example.com/bg.jpg" --resolution "720P"

Options

• --prompt: Editing instruction (required, max 5000 non-Chinese chars)
• --video: Input video URL to edit (required, mp4/mov, 3-60s input, output max 15s)
• --reference-images: Reference image URLs for editing (optional, 0-5 images)
• --resolution: Output resolution - 720P or 1080P (default: 1080P)
• --audio-setting: Audio handling - auto (default) or origin (keep original audio)
• --no-watermark: Disable "Happy Horse" watermark (default: enabled)
• --seed: Random seed for reproducibility [0, 2147483647]

Note: Output aspect ratio and duration follow the input video. No --ratio or --duration parameters. If input video > 15s, system auto-truncates to first 15s.

video-edit tasks-get (round-robin)

python3 {baseDir}/scripts/happyhorse-magic.py video-edit-get --task-id "<TASK_ID_FROM_VIDEO_EDIT>"

Key Differences from Wan 2.7

Feature	HappyHorse	Wan 2.7
Watermark default	Enabled ("Happy Horse")	Disabled ("AI Generated")
prompt_extend	Not supported	Supported
negative_prompt	Not supported	Supported
seed parameter	Supported (all models)	Supported (all models)
I2V modes	First-frame only	First-frame, First+Last-frame, Video continuation
R2V reference type	Images only (character1/2)	Videos + Images (视频1/图片1)
R2V reference count	1-9 images	Max 3 videos + 5 images
Video Edit input	3-60s, output max 15s	2-10s
Video Edit prompt	Required	Optional

Important Notes

• Task ID Validity: Task IDs are valid for 24 hours
• Video URL Validity: Generated video URLs are valid for 24 hours, download immediately
• Content Review: Both input and output are subject to content safety review
• Billing: Based on resolution (1080P > 720P) × duration (seconds)
• Watermark: Default enabled with "Happy Horse" text; use --no-watermark to disable