generate-image

generate-image Skill

Trigger Conditions

Invoke this skill when:

• User types /generate-image [prompt] — optionally with --model [model-id]
• User says "generate an image of X", "create a picture of X", "draw X", "make a visual of X", "show me an image of X"
• The task clearly requires a visual asset (logo, banner, illustration, diagram, etc.)

How to Execute

The generate-image script path was provided at session start (see [image-generator] Tool script ready in session context).

Use the Bash tool to call it directly:

node "<script-path-from-session-context>" "<prompt>" [--model <model-id>] [--output-dir <dir>]

The script handles everything: API call, response parsing, base64 decode, file save. It prints Image saved: <path> on success or a clear error message on failure.

Examples

# Default model (google/gemini-3.1-flash-image-preview)
node "/path/to/scripts/generate.js" "a sunset over mountains"

# Override model
node "/path/to/scripts/generate.js" "a minimalist cat logo" --model google/imagen-4.0-generate-001

# Save to a specific directory
node "/path/to/scripts/generate.js" "abstract art" --output-dir ./assets

After the script runs

Report to the user:

• The saved file path
• The model used
• Optionally, offer to open or display the image

Model Reference

All models use OpenRouter's /api/v1/chat/completions endpoint — no special routing needed.

Model	Notes
google/gemini-3.1-flash-image-preview	Default — best quality
google/imagen-4.0-generate-001	High-fidelity photorealism
black-forest-labs/flux-kontext-pro	Fast, artistic styles