Video Breakdown

Provides quantitative quality scores and detailed shot-by-shot breakdowns for videos to evaluate technical quality and narrative structure.

Security Scan

The skill claims to analyze videos via a third‑party backend but the instructions omit crucial implementation details (endpoint, auth, privacy), so its behavior and data flows are unclear and warrant caution.

Purpose & Capability

The declared purpose (shot-by-shot and quality analysis) matches the instructions (submit video_url and poll a backend). However the SKILL.md references a "custom backend service" (and fal.ai) but provides no base URL, host, or authentication details — leaving an unexplained gap between claimed capability and how it would actually run.

Instruction Scope

Runtime instructions tell the agent to POST /analyze and poll /status/{task_id}, but they do not specify the full service endpoint, authentication, or policies for handling submitted video URLs. That makes the instructions incomplete and also raises privacy concerns because user-supplied video URLs would be sent to an unspecified external service.

Install Mechanism

No install spec and no code files are present (instruction-only), so nothing will be downloaded or written to disk by the skill itself. This is the lowest-risk install pattern, but also increases reliance on external platform tooling to implement the described API calls.

Credentials

The skill requests no environment variables or credentials. That is consistent with an anonymous public API, but inconsistent with the SKILL.md mention of a third-party backend (fal.ai) — such backends often require API keys. The absence of declared auth is an unexplained omission.

Persistence & Privilege

The skill does not request persistent or elevated privileges (always: false), and does not attempt to modify other skills or system configuration. Autonomous invocation is allowed by platform defaults but is not combined with additional privileges here.

What to consider before installing

This skill describes a useful video-analysis service but leaves important questions unanswered. Before installing or using it, ask the provider for: (1) the full backend URL(s) the agent will call, (2) authentication requirements (API key or token) and why no env vars are declared, (3) a privacy/data retention policy (will your videos be stored, for how long, used for model training?), (4) whether private or internal URLs will be accessed or if requests are proxied, and (5) an auditable implementation (code or platform-integrated tool) you can review. If you must test it, only submit non-sensitive/public videos and monitor outbound network activity; avoid sending proprietary or private media until you have explicit assurances about storage, access control, and the third‑party operator's identity.

An advanced video analysis skill that provides both high-level quality assessment and detailed shot-by-shot breakdowns. It helps content creators, editors, and marketers to objectively evaluate video quality and deconstruct narrative structures. It helps content creators, editors, and marketers to objectively evaluate video quality and deconstruct narrative structures.

This skill is powered by a custom backend service that leverages state-of-the-art video understanding models from fal.ai.

Core Capabilities

Capability	Description	Use Case
Quality Critique	Provides a quantitative score (1-10) and qualitative comments for key technical aspects of a video.	Quickly evaluate if user-generated content meets quality standards; compare different versions of a video.
Shot Breakdown	Delivers a detailed breakdown of each shot, including timestamps, visual descriptions, and camera work.	Deeply analyze a competitor's video; deconstruct a film scene for educational purposes; create a highlight reel script.

How It Works

This skill acts as a client to a dedicated backend service. The agent submits a video URL and an analysis_type to the service, which then queues the job and returns a task_id. The agent can then poll the status of the job until it is completed.

Workflow

1. Agent: Calls the /analyze endpoint with video_url and analysis_type.
2. Service: Submits the job to the AI model and returns a task_id.
3. Agent: Periodically calls the /status/{task_id} endpoint.
4. Service: Returns the job status (queued, in_progress, or completed).
5. Agent: Once completed, retrieves and presents the JSON result to the user.

Usage

1. Quality Assessment

Goal: Get a technical quality report for a video.

Agent Action:

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

Expected Output (after polling):

{
  "resolution": {
    "score": 9,
    "comment": "The resolution is high, and the main subject is exceptionally sharp..."
  },
  "lighting": {
    "score": 9,
    "comment": "The lighting is excellent, effectively utilizing vibrant neon..."
  },
  "audio": {
    "score": 7,
    "comment": "Audio is clear, but some background noise is present..."
  },
  "stability": {
    "score": 10,
    "comment": "The video exhibits outstanding stability..."
  }
}

2. Shot-by-Shot Analysis (拉片)

Goal: Get a detailed scene breakdown of a video.

Agent Action:

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

Expected Output (after polling):

[
  {
    "start_time": "00:00",
    "end_time": "00:04",
    "description": "A stylish young woman is walking confidently towards the camera...",
    "camera_work": "The camera is static and appears to be at a medium-low angle..."
  },
  {
    "start_time": "00:04",
    "end_time": "00:08",
    "description": "Close-up on the woman's face as she smiles subtly...",
    "camera_work": "The camera zooms in slowly, focusing on her expression..."
  }
]

Backend Service API Reference

• POST /analyze: Submits a new analysis job.
  • Body: {"video_url": "string", "analysis_type": "string"}
• GET /status/{task_id}: Checks the status of a job.
• GET /analysis_types: Lists available analysis_type values.