Fix size validation for OpenAI model in Art skill

[Azd325]

What Problem This Solves

The Art skill's Generate.ts tool was rejecting valid OpenAI image sizes (1536x1024, 1024x1536) with an "Invalid size" error.

Root Cause

The size validation logic was falling through to check OpenAI sizes against Replicate's aspect ratio format, causing valid sizes to be rejected.

The Fix

Changed the validation from:

if (parsed.model === "gpt-image-1" && !OPENAI_SIZES.includes(...))

To proper if-else-if structure:

if (parsed.model === "gpt-image-1") {
  if (!OPENAI_SIZES.includes(...)) { throw error }
}

Testing Done

✅ Verified OpenAI model now works with all valid sizes: 1024x1024, 1536x1024, 1024x1536
✅ Tested image generation completes successfully
✅ Confirmed no impact on other models (nano-banana-pro, flux, nano-banana)

Example

bun run Generate.ts --model gpt-image-1 --size 1536x1024 --prompt "test" --output test.png
# Previously: Error: Invalid size: 1536x1024
# Now: Successfully generates image

[danielmiessler]

Thank you @Azd325 for this Art skill fix! 🙏

Great catch on the size validation logic. We've released PAI v2.1 which renamed all packs from kai-* to pai-*, so this path no longer exists.

If you'd like to re-apply this fix to Packs/pai-art-skill/src/skills/Art/Tools/Generate.ts, we'd happily merge it!

See the release: https://github.com/danielmiessler/PAI/releases/tag/v2.1.0

[Azd325]

Updated the fix to apply to the new path. New PR #348