rajshah4/skills/video-visualizer

Video Visualizer Workflow (AI Engineer Aesthetic)

Overview

This skill provides a structured workflow for an agent to act as a visual model. It analyzes video scripts and source documents to propose clean, highly technical, and minimalist background visuals. The target audience is practical AI engineers.

Design Philosophy (The "Anti-Glow" Manifesto)

• Target Audience: Practical AI Engineers.
• Aesthetic: Minimalist, clean, schematic. Think academic paper figures, Stripe/Vercel documentation, or Edward Tufte's data-ink ratio.
• Rules:
  • NO glowing shapes, NO neon, NO crystals.
  • Use stark contrasts (e.g., crisp text vs. blurred text) to show focus.
  • Use technical examples (JSON, code snippets, architecture diagrams) instead of generic metaphors.
  • Keep it grounded in reality (e.g., standard RL flowcharts, context window bounding boxes).

Workflow Steps

Step 1: Identification & Proposal

1. Analyze Source Material: Review the research paper/article for existing figures.
2. Analyze Script Narrative: Map out the script's core concepts.
3. Draft Proposal: Propose schematic, technical visuals (e.g., text swath pruning, pipeline flowcharts).
4. User Confirmation: Present the visual ideas to the user for feedback.

Step 2: Prompt Engineering

1. Translate to Prompts: Convert approved ideas into highly specific image generation prompts.
2. Set Style Parameters: Enforce the design philosophy (e.g., "flat vector diagram", "monochromatic with one accent color", "clean lines", "technical documentation style").

Step 3: Generation (API Invocation)

1. Request API Key: Obtain the key for the image generator (e.g., Nano Banana 2).
2. Invoke API: Use the generated prompts to call the endpoint.
3. Deliver Assets: Provide the generated visuals to the user.

Usage Trigger

Use this skill when the user asks for help conceptualizing visuals for an AI/technical script or generating background graphics tailored for engineers.