feat(art): Add video generation with Google Veo models

[b1ackmartian]

Summary

Closes #420

This PR extends the Art skill with AI video generation capabilities using Google's Veo models, enabling animated content with the same Excalidraw hand-drawn aesthetic used for images.

Changes

Generate.ts

• Added Veo model support: veo-3.1, veo-3.1-fast, veo-3, veo-3-fast, veo-2
• New video generation functions with async polling
• Video extension for longer content (up to ~148 seconds)
• Image-to-video (animate existing images with --first-frame)
• Cost estimation with --dry-run
• Optional AI-generated audio

New CLI Flags

Flag	Description
--duration <sec>	Target video duration (default: 8)
--first-frame <path>	Image for first frame
--last-frame <path>	Image for last frame (Veo 3.1)
--negative-prompt <text>	Elements to exclude
--audio / --no-audio	Enable/disable AI audio
--resolution <res>	720p, 1080p, or 4K
--extend	Extend existing video
--input-video <path>	Video to extend
--dry-run	Estimate cost

New Files

• Workflows/Video.md - Comprehensive video generation workflow

Updated Files

• SKILL.md - Added video routing and examples
• README.md - Updated version, added video examples

Pricing

Model	Cost/Second	8s Video
veo-3.1-fast	$0.15	$1.20
veo-3.1	$0.40	$3.20
veo-3	$0.40	$3.20
veo-2	$0.35	$2.80

Example Usage

# Quick 8-second draft
bun run Generate.ts --model veo-3.1-fast --prompt "..." --output ~/Downloads/video.mp4

# Animate existing image
bun run Generate.ts --model veo-3.1 --prompt "..." --first-frame image.png

# Production with audio
bun run Generate.ts --model veo-3.1 --prompt "..." --audio

# Cost estimation before generating
bun run Generate.ts --model veo-3.1 --prompt "..." --duration 30 --dry-run

Test Plan

• Verify basic video generation with veo-3.1-fast
• Test --audio and --no-audio flags
• Verify image-to-video with --first-frame
• Test video extension with --duration > 8
• Confirm --dry-run cost estimation accuracy
• Validate error handling for missing API key

🤖 Generated with Claude Code