video-agent（视频 Agent）

Video Agent - AI Content Generation Suite

A comprehensive AI content generation package providing a unified interface across 35+ models for image, video, and audio creation.

When to Use This Skill

• Text-to-image generation
• Image-to-image transformations
• Text-to-video creation
• Image-to-video animation
• Professional text-to-speech
• Multi-step content pipelines
• Batch content generation

Supported Providers

FAL AI

• FLUX models (text-to-image)
• Image transformations
• Fast inference

Google Vertex AI

• Imagen 4 (text-to-image)
• Veo (text-to-video)
• High quality outputs

ElevenLabs

• 20+ voice options
• Professional TTS
• Multiple languages

OpenRouter

• Access to various LLMs
• Text generation
• Content writing

Core Capabilities

Image Generation

Generate image:
Prompt: "A serene Japanese garden at sunset"
Model: flux-pro
Size: 1024x1024
Style: photorealistic

Available Models:

• FLUX Pro/Dev (FAL)
• Imagen 4 (Google)
• Stable Diffusion variants

Video Creation

Generate video:
Prompt: "Ocean waves crashing on rocky shore"
Model: veo
Duration: 5 seconds
Resolution: 1080p

Available Models:

• Google Veo
• MiniMax Hailuo
• Kling

Image-to-Video

Animate image:
Source: /path/to/image.png
Motion: "gentle zoom out with particle effects"
Duration: 4 seconds

Text-to-Speech

Generate audio:
Text: "Welcome to our product demo..."
Voice: professional-female-1
Speed: 1.0
Output: welcome.mp3

Voice Options:

• Professional male/female
• Casual conversational
• Narrator styles
• Multiple accents

Pipeline Orchestration

YAML Configuration

pipeline: product-demo
steps:
  - name: generate-logo
    type: image
    model: flux-pro
    prompt: "Modern tech logo for AI startup"

  - name: create-intro
    type: video
    model: veo
    prompt: "Logo animation reveal"

  - name: add-voiceover
    type: audio
    model: elevenlabs
    text: "Introducing the future of AI..."
    voice: professional-male

  - name: combine
    type: merge
    inputs: [create-intro, add-voiceover]

JSON Configuration

{
  "pipeline": "social-content",
  "parallel": true,
  "steps": [
    {
      "type": "image",
      "variants": 4,
      "prompt": "Product hero shot"
    }
  ]
}

Cost Management

Real-time Estimation

Estimate cost for:
- 10 images (1024x1024)
- 2 videos (5 seconds)
- 1 audio (60 seconds)

Estimated: $2.45

Budget Limits

budget:
  max_per_job: $5.00
  max_daily: $50.00
  alert_threshold: 80%

Performance Features

Parallel Execution

Generate 10 image variants in parallel
Threads: 4
Expected speedup: 2-3x

Caching

• Automatic prompt caching
• Reuse similar generations
• Reduce redundant API calls

CLI Commands

# Image generation
video-agent image "prompt" --model flux-pro --size 1024

# Video generation
video-agent video "prompt" --model veo --duration 5

# Audio generation
video-agent audio "text" --voice professional-female

# Pipeline execution
video-agent pipeline config.yaml

# Cost check
video-agent cost --estimate

Python API

from video_agent import ImageGenerator, VideoGenerator

# Generate image
img = ImageGenerator(model="flux-pro")
result = img.generate("sunset over mountains")

# Generate video
vid = VideoGenerator(model="veo")
result = vid.generate("timelapse of clouds")

Setup

1. Install Package

pip install video-agent-claude-skill

2. Configure API Keys

export FAL_API_KEY="your-key"
export GOOGLE_VERTEX_KEY="your-key"
export ELEVENLABS_API_KEY="your-key"

3. Verify Setup

video-agent status

Use Cases

• Marketing: Product images, promo videos
• Social Media: Content at scale
• Education: Explainer videos, voiceovers
• Prototyping: Visual concepts, mockups
• Automation: Batch content pipelines

Credits

Created by donghaozhang. Licensed under MIT.