Gpt Image 2 Skill

GPT Image - Claude Code Skill

Claude can't reliably call OpenAI's image API from memory. The surface keeps moving, the size constraints are easy to get wrong, and more often than not Claude either guesses the signature or silently skips "generate me an image" tasks entirely. This skill fixes that.

Two CLI scripts wrapped as a [Claude Code](https://claude.com/claude-code) skill: `generate.py` for text-to-image with `gpt-image-2`, and `edit.py` for editing or compositing up to 10 reference images. Dependencies install on first run via `uv` and PEP 723 inline metadata. No glue code. No boilerplate.

Requirements

Requirement	Purpose
uv	Runs the scripts and manages dependencies automatically. No pip install needed.
OpenAI API key	Must have access to gpt-image-2. Requires API organization verification.
Claude Code (optional)	Only needed if you want Claude to invoke the skill automatically. The scripts also work standalone.

Installing uv

`uv` is a fast Python package manager from Astral. It replaces `pip` + `venv` + `pyenv` with a single tool that's roughly 10-100x faster. These scripts rely on `uv run` to fetch dependencies on first invocation, so it must be installed.

**macOS / Linux:**

curl -LsSf https://astral.sh/uv/install.sh | sh

**Windows (PowerShell):**

powershell -c "irm https://astral.sh/uv/install.ps1 | iex"

**Homebrew:**

brew install uv

**Verify the install:**

uv --version

If `uv` isn't on your PATH after install, restart your shell or follow the on-screen instructions from the installer. Full docs: .

Setup

**1. Clone or copy the skill into your Claude Code skills directory.**

# User-wide (recommended)
git clone https://github.com/dshark3y/gpt-image-2-skill.git ~/.