imagegen

Generate and edit raster images using the built-in image_gen tool or a GPT Image CLI fallback.

Updated Aug 4, 2026
One-click install
npx skills add https://github.com/sjc786526-coder/RONDO --skill imagegen-sjc786526-coder
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: imagegen
Source: https://github.com/sjc786526-coder/RONDO/tree/main/multidev/codex-rs/skills/src/assets/samples/imagegen
Command: npx skills add https://github.com/sjc786526-coder/RONDO --skill imagegen-sjc786526-coder

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires openai, pillow, and includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve? Creating bitmap assets such as photos, illustrations, sprites, mockups, and transparent cutouts normally requires leaving the coding workflow for a separate image tool. This Skill lets Codex generate or edit raster images directly in the project context, with a built-in tool path and a guarded CLI fallback for advanced API controls. ## Core Features & Use Cases - Built-in generation and editing: Uses the built-in image_gen tool by default for new images, reference-guided generation, and edits, with no API key required. - Transparent-background workflow: Generates subjects on a flat chroma-key background and removes it locally with remove_chroma_key.py, escalating to CLI gpt-image-1.5 true transparency only after explicit user confirmation. - Fallback CLI with batch support: scripts/image_gen.py exposes generate, edit, and generate-batch subcommands with model, quality, size, mask, and output-path controls when the user explicitly opts in. - Use Case: A developer building a landing page asks for a hero product image; the Skill generates it with the built-in tool, validates the result, and copies the final PNG into the project workspace with a versioned filename. ## Quick Start Ask Codex to generate a hero image of your product for the landing page and save it into the project assets folder.

Frequently Asked Questions about imagegen

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate an image for my project with Codex?▼

Describe the image you need and its intended use, and the Skill uses the built-in image_gen tool by default. The finished asset is copied into your workspace with a versioned filename rather than left in the default generated-images directory.

How do I create an image with a transparent background?▼

The default path generates the subject on a flat chroma-key background, then removes it locally with the remove_chroma_key.py helper to produce an alpha PNG. True model-native transparency requires the CLI fallback with gpt-image-1.5 and is used only after you confirm.

When should I use the image_gen.py CLI instead of the built-in tool?▼

Use the CLI only when you explicitly need API-level controls such as model selection, quality, masks, output formats, or JSONL batch generation. It requires OPENAI_API_KEY and network access, and the Skill never switches to it silently.

Does gpt-image-2 support transparent backgrounds?▼

No, gpt-image-2 does not support the background=transparent parameter. Transparent output with the CLI requires gpt-image-1.5 with output format png or webp, which is why the default workflow uses chroma-key generation plus local removal.

When should I not use AI image generation for an asset?▼

Avoid it when the task is better served by editing existing SVG or vector assets, extending an established icon or logo system, or building the visual directly in HTML, CSS, or canvas. The Skill is intended for bitmap output, not code-native graphics.