gpt-image

name	gpt-image
description	Generate or edit images with gpt-image-2. Use when the user wants image generation or image editing via Yunwu first, with Right Code as fallback.

GPT Image

Use this skill for gpt-image-2 image generation and editing.

The agent should not rely on bundled code. Construct the request on demand with curl, Python, JavaScript, or another shell-friendly client.

Source docs

• OpenAI image generation guide: https://developers.openai.com/api/docs/guides/image-generation
• OpenAI image generation reference: https://developers.openai.com/api/reference/resources/images/methods/generate/
• OpenAI gpt-image-2 model page: https://developers.openai.com/api/docs/models/gpt-image-2
• OpenAI prompting guide: https://developers.openai.com/cookbook/examples/multimodal/image-gen-models-prompting-guide
• Right Code draw overview: https://docs.right.codes/docs/rc_extension/draw/
• Right Code image generation page: https://docs.right.codes/docs/rc_extension/draw/images-generations.html

Backends

Yunwu (default and preferred)

• Base URL: https://yunwu.ai/v1
• API key env var: YUNWU_API_KEY
• Model: gpt-image-2
• Use this backend first unless the user explicitly asks otherwise.

Right Code (fallback)

• Base URL: https://www.right.codes/draw
• API key env var: RC_API_KEY
• Model: gpt-image-2
• Use this only if Yunwu fails or is unavailable.
• Assume text-only generation here; do not rely on image upload or editing support.

Call shape

Use the OpenAI Images API shape when the backend supports it:

• Text-to-image: POST /images/generations
• Edit image: POST /images/edits
• Model: gpt-image-2
• Prompt: plain natural language, from the current request only
• Default quality: high

Right Code documents a simpler draw endpoint:

• POST /v1/images/generations
• Fields: model, prompt, optional image, optional size, optional response_format
• Response: data[0].url

Example: Yunwu text-to-image

curl -sS "https://yunwu.ai/v1/images/generations" \
  -H "Authorization: Bearer $YUNWU_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-image-2",
    "prompt": "A studio product photo of a black ceramic mug on a warm oak table",
    "size": "1024x1024",
    "quality": "high"
  }'

Example: Right Code text-to-image

curl -sS "https://www.right.codes/draw/v1/images/generations" \
  -H "Authorization: Bearer $RC_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-image-2",
    "prompt": "A studio product photo of a black ceramic mug on a warm oak table",
    "size": "1024x1024",
    "quality": "high",
    "response_format": "url"
  }'

Caveats

• Prefer Yunwu first; Right Code is a fallback.
• gpt-image-2 does not support transparent backgrounds.
• Do not send input_fidelity for gpt-image-2.
• Use quality: high by default unless the user explicitly wants a faster draft.
• Keep sizes within the documented constraints: both edges divisible by 16, aspect ratio no wider than 3:1, and treat outputs above 2560x1440 as experimental.
• Right Code returns image URLs in data[0].url; do not assume base64 there.
• Right Code should be treated as text-to-image only in this skill.
• If you need edits, use Yunwu and verify the backend accepts the OpenAI multipart edit shape before using local files.
• If the request includes reference images, route the work to Yunwu and be explicit about which image is the subject, palette, background, or composition reference.

Prompting

• Describe the scene in natural language.
• State the subject, style, lighting, framing, and intended use.
• For text in images, specify the exact text and where it should appear.
• For edits, say what must stay unchanged and what must change.