version: "1.0.0" name: image-gen description: AI image generation with OpenAI, Google and DashScope APIs. Supports text-to-image, reference images, aspect ratios, and direct Cloudflare R2 upload. Sequential by default; parallel generation available on request. Use when user asks to generate, create, or draw images.

Image Generation

Generate images via OpenAI, Google Gemini, or DashScope APIs. R2 upload is the default workflow for persistent article/social images.

Scripts

SKILL_DIR/
  scripts/main.ts        — generate image (+ optional R2 upload)
  scripts/upload-r2.ts   — upload any existing file to R2
  scripts/r2.ts          — R2 upload module (AWS SigV4, no external deps)
  scripts/providers/     — google.ts / openai.ts / dashscope.ts
  scripts/types.ts       — shared types

Usage

SKILL_DIR="/path/to/image-gen"
 
# Basic generation
npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "A cat" --image cat.png
 
# With aspect ratio
npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "A landscape" --image out.png --ar 16:9
 
# From prompt file
npx -y bun ${SKILL_DIR}/scripts/main.ts --promptfiles prompt.md --image out.png
 
# Generate + upload to R2 (prints public URL)
npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "A cat" --image cat.png --r2
 
# Generate + upload with custom R2 key
npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "A cat" --image cat.png --r2 --r2-key articles/hero.png
 
# Upload existing file to R2
npx -y bun ${SKILL_DIR}/scripts/upload-r2.ts --file image.png
npx -y bun ${SKILL_DIR}/scripts/upload-r2.ts --file image.png --key custom/path.png
 
# With reference image
npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "Make it blue" --image out.png --ref source.png
 
# Force provider
npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "A cat" --image out.png --provider openai

Options

Environment Variables

Set in the vault-local .env.r2, ~/.skills/.env, or ~/.baoyu-skills/.env:

Provider Selection

Auto-detects from available API keys. Priority: Google → OpenAI → DashScope. With --ref: Google first (Gemini multimodal), then OpenAI.

R2 Upload

Uses AWS SigV4 signing with built-in Node.js crypto — no external dependencies. With --r2, prints the R2 public URL instead of the local path. For article/social images, always use --r2 and insert the returned URL into Markdown; keep local files only as temporary cache.

Generation Mode

Sequential by default. For parallel batch generation, use multiple background subagents.

Default Workflow in This Vault

For any image that will be referenced by articles, tweets, covers, or long-lived notes:

1. Generate the image to a temporary or ignored local path.
2. Upload immediately with --r2 or scripts/upload-r2.ts.
3. Insert the returned R2_PUBLIC_BASE_URL URL into Markdown.
4. Do not rely on committed local image files as the canonical asset.