Video Breakdown

A professional video analysis skill powered by a dual-model architecture: ByteDance Seed-2.0-Mini for rapid previews and Google Gemini 2.5 Pro for deep, cinematic-grade analysis. It provides quantitative quality assessments and meticulous shot-by-shot breakdowns (拉片) for content creators, editors, and filmmakers.

Core Capabilities

Model Selection

This skill uses two models, selectable via the model parameter:

How It Works

The skill calls a hosted proxy service that routes requests to OpenRouter, which dispatches to the selected model. The response is synchronous — the full analysis result is returned directly in the API response.

Workflow

1. Agent: Calls POST /api/analyze with video_url, analysis_type, and optionally model.
2. Proxy: Forwards the request to OpenRouter with the selected model.
3. Model: Analyzes the video and returns structured JSON.
4. Agent: Presents the parsed result to the user.

Usage

1. Quick Quality Assessment (Seed-2.0-Mini)

Goal: Get a fast quality report for a video.

Agent Action:

{
  "tool": "video-breakdown.analyze",
  "args": {
    "video_url": "https://example.com/my-video.mp4",
    "analysis_type": "quality_critique",
    "model": "quick"
  }
}

2. Deep Shot-by-Shot Analysis (Gemini 2.5 Pro)

Goal: Get a professional, frame-accurate shot breakdown.

Agent Action:

{
  "tool": "video-breakdown.analyze",
  "args": {
    "video_url": "https://example.com/scene.mp4",
    "analysis_type": "shot_breakdown",
    "model": "full"
  }
}

Expected Output:

[
  {
    "shot_number": 1,
    "start_time": "00:00",
    "end_time": "00:04",
    "duration_seconds": 4,
    "shot_type": "Medium Shot",
    "camera_movement": "Static",
    "subject": "Young woman walking toward camera",
    "action": "Subject walks confidently, making direct eye contact",
    "narrative_function": "Establishes protagonist and sets confident tone",
    "audio_notes": "Upbeat music begins, no dialogue"
  }
]

3. Content Strategy Analysis

Goal: Evaluate a video's social media performance potential.

Agent Action:

{
  "tool": "video-breakdown.analyze",
  "args": {
    "video_url": "https://example.com/reel.mp4",
    "analysis_type": "content_strategy",
    "model": "full"
  }
}

Backend Service API Reference

The proxy service is deployed on Vercel Pro (300s timeout).

POST /api/analyze

Submits a video for analysis.

Request Body:

{
  "video_url": "string (required)",
  "analysis_type": "quality_critique | shot_breakdown | content_strategy (required)",
  "model": "quick | full (optional, default: full)"
}

Response:

{
  "model_used": "google/gemini-2.5-pro",
  "analysis_type": "shot_breakdown",
  "result": { ... }
}

GET /api/health

Returns service status and available models.

Deployment

The proxy service requires one environment variable:

OPENROUTER_API_KEY=<your-openrouter-api-key>

Deploy to Vercel from the proxy/ directory within this skill.