ChatGPT Image 2.5 Claude Skill — Development skill for Claude Code
A skill that generates and edits images with ChatGPT Images 2.5, using your own ChatGPT login through the Codex CLI.
How to install ChatGPT Image 2.5 Claude Skill
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open AmineTabtani/chatgpt-Image-2.5-claude-skill and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What ChatGPT Image 2.5 Claude Skill does
A skill that generates and edits images with ChatGPT Images 2.5, using your own ChatGPT login through the Codex CLI. No OpenAI API key needed. Unofficial community project. Not affiliated with, endorsed by, or sponsored by OpenAI or Anthropic.
Alternatives in Development
- Imagegen — Generate and edit images using OpenAI's Image API for projects 14.6k ★
- Codexbar — Show usage stats for OpenAI Codex and Claude Code, without having to login 8.8k ★
- Opencode Claude Auth — OpenCode plugin that uses your existing Claude Code credentials — no separate login needed 903 ★
README
chatgpt-images
A [Claude Code](https://claude.com/claude-code) skill that generates and edits images with **ChatGPT Images 2.5**, using your own ChatGPT login through the Codex CLI. No OpenAI API key needed.
**Unofficial community project.** Not affiliated with, endorsed by, or sponsored by OpenAI or Anthropic. ChatGPT and Codex are trademarks of OpenAI; Claude and Claude Code are trademarks of Anthropic.
What it does
Once installed, Claude Code uses the skill whenever you ask for a bitmap image: game sprites, icons, textures, UI art, concept art, backgrounds, logos, mockups, illustrations or photos. It can also edit an existing image and make cutouts on a transparent background.
Example prompts in Claude Code:
Make a 2D pixel-art health potion icon with a transparent background and save it to assets/potion.png
Edit hero.png: make the sky a warm sunset, keep the character and framing unchanged.
Under the hood, Claude runs `chatgpt-images/scripts/codex-image.mjs`, which:
- runs
codex execnon-interactively in a read-only sandbox and asks Codex's built-in$imagegentool to generate or edit the image; - finds the result in
~/.codex/generated_images/and copies it to the path you asked for (adding-v2,-v3... instead of overwriting); - prints a JSON summary (paths, dimensions, whether the image has transparency) that Claude checks before reporting back.
A run usually takes from about 30 seconds to a few minutes. Square images come out around 1254x1254.
**Usage limits:** every image is generated with your ChatGPT plan and counts against its Codex usage limits. When the limit is reached, generation fails until it resets.
Requirements
- Claude Code
- Node.js 18 or newer (
node --version) - Codex CLI 0.159 or newer, signed in with a ChatGPT account (not an API key):
If you are not signed in, run `codecodex --version codex login status # should say "Logged in using ChatGPT"
Related Skills
Pvc Img Gen Codex
Claude Code skill that makes and edits real images through the Codex CLI from the OpenAI ChatGPT VS Code exten
Claude Code Image Gen
A Claude Code skill that generates images through OpenAI, Gemini, Higgsfield, FLUX, or any OpenAI-compatible e
Gs Admin CLI Docs
Unofficial community docs, knowledge base, and Claude Code plugin for the Gainsight Admin CLI (gs-admin). Not
Tldraw Diagram Skill
Claude Code skill for generating tldraw diagrams (.tldr files & SDK code) with a headless PNG/SVG renderer. Un
Edit Image
This command edits existing images based on provided edit prompts using Replicate's API directly via curl.
Codex Imagegen Skill
A Claude Code skill that generates images via Codex CLI's $imagegen, using your ChatGPT subscription instead o
Related Agents
Image Generator
Generates and edits images using the OpenAI Image API via the bundled CLI (scripts/image_gen.py). Use for conc
Image Craft Expert
Crafts optimized text-to-image prompts and generates images using Gemini (nano-banana) and/or ChatGPT (gpt-ima
Codex Agent
Runs OpenAI Codex on the user's ChatGPT subscription. Three modes - consult (OS-sandboxed read-only analysis,