SkillFoxx

gpt-image-bridge skill

A skill that adds image generation to Claude Code and Cursor through the Codex CLI on a ChatGPT subscription, no API key.

  • 84 ★ on the repository
  • Last commit
  • MIT
Skill

Extended access

Why this level

  • The wrapper runs `codex exec` with write access to a work and a temp folder
  • It runs on your ChatGPT subscription, and automated use may conflict with OpenAI terms
All reasons and checks

oakplank/gpt-image-bridgeAuthor:oakplank

Install

In your terminal, with SkillFoxx CLI

npx skillfoxx add skills/gpt-image-bridge

Detects the agents on your machine, shows the access level and pins the version.

Other ways to install

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • The wrapper runs `codex exec` with write access to a work and a temp folder
  • It runs on your ChatGPT subscription, and automated use may conflict with OpenAI terms

This is third-party code. Review the repository files before installing.

What it does

A thin bash wrapper calls `codex exec` and its `image_generation` tool and copies the finished PNG to the path the agent gave. The script is named `gpt-image-2` and takes a prompt, an absolute output path and an optional `--size`. The agent then reads the image back into context and checks the result. The `install.sh` installer puts the skill in `~/.claude/skills/gpt-image-bridge/`, and Claude Code reaches for it on its own when asked for a mockup, logo, avatar or illustration. For other agents the script on your `PATH` and a few lines in `AGENTS.md` or `.cursorrules` are enough. It pairs with design skills: when they need a picture, they call this wrapper. Codex picks the image model, and the wrapper does not pin one.

Who it is for. ChatGPT Plus, Pro or Team subscribers working in Claude Code or Cursor who want images without an API key.

Good fit when

  • A design skill follows an image-first workflow and needs a mockup or reference
  • A landing page needs a hero image, illustration or avatar and you have no OpenAI API key
  • An agent in Cursor or aider must get a PNG from a shell command and a ready prompt
  • The result has to be judged: the skill returns the image to context and the agent refines the prompt itself

Not a fit when

  • You need bulk generation with predictable speed: one call takes 4 to 6 minutes
  • You need a simple picture such as a solid placeholder or a chart: it is easier to draw with Python code
  • You need pixel-level edits to an existing image: the wrapper creates new images

Example request

Generate a hero image for a coffee page with warm light and a dense composition, save it to `/tmp/hero.png` and show me the result.

Limitations

The `codex` CLI must be installed and signed in with `codex login` on a ChatGPT subscription. Usage draws on the subscription message quota and heavy use can hit the cap. Automating a consumer subscription ultimately depends on OpenAI terms, so the author warns that the decision to use it this way is yours. One call takes a long time because Codex reasons first, so Claude Code needs the maximum shell timeout. Windows needs Git Bash or WSL.

How to disable. Delete the `~/.claude/skills/gpt-image-bridge/` folder. If you symlinked the script into `/usr/local/bin`, remove that too. Signing out of Codex is a separate step.

FAQ

How do I install the skill in Claude Code?

Clone the repository and run `./install.sh`. The installer copies the skill into `~/.claude/skills/gpt-image-bridge/` and needs no further setup. Beforehand `codex login status` must show a ChatGPT login.

Can I use it in Cursor or aider?

Yes. The script is self-contained: make it executable, put it on your `PATH` and add lines with the `gpt-image-2` command and how to call it to `AGENTS.md`, `.cursorrules` or the system prompt.

Why does an image take about five minutes?

Codex reasons and refines the prompt before the image tool fires. Speed depends on the `reasoning_effort` setting in Codex, and Claude Code needs the maximum timeout for the call.

How is this different from calling the Images API directly?

The API needs a key and charges per image but is faster and passes the prompt verbatim. Here usage comes from the subscription quota and the Codex model refines the prompt.

What if the call fails?

The wrapper prints the tail of the Codex log to stderr. Make sure `codex` is installed and logged in and that `image_generation` is enabled via `codex features list`.

Discussion

No questions or reviews yet.

Skills

A skill and CLI for generating and editing images with GPT Image 2 and 2.5 plus a library of ready prompts

SkillExtended access6.1K
Needs a VPN

A skill teaching agents to work with GPT Image 2. OpenAI API access is needed and it is unavailable from Russia without workarounds

SkillExtended access87

A Rust skill and CLI for agents to generate and edit images with GPT Image 2, make transparent PNGs and answer in JSON.

SkillExtended access147

A plugin that draws and edits images in Claude Code through the built-in image_gen of local Codex on a ChatGPT login, no API key.

PluginExtended access126

A Claude Code skill that compiles a vague request into a finished gpt-image-2 prompt and checks it with a validator script

SkillExtended access331

Skill, MCP server and CLI that let any agent call Codex image generation, after picking a style from several directions

SkillExtended access136

Comparison with similar tools

Cards that share the main search query, and how they differ.

ToolOfficialRepository starsMain difference
gpt-image-bridge (this card)no84
Claude Image (GPT Image 2)no87teaches agents to work with GPT Image 2 through an OpenAI API key rather than a subscription
GPT-Image2-Skillno6,086a skill and CLI for generating and editing GPT Image 2 and 2.5 images with a prompt library

README in short

The page explains the flow: the agent calls the wrapper, which runs `codex exec`, Codex writes a file into a temp folder and the wrapper copies it to the target path. Requirements, installs for Claude Code and other agents, a block to paste to an agent, a comparison with the direct API, caveats and the MIT license follow.

README badge

Original SKILL.md

---
name: gpt-image-bridge
description: Use when the user asks for an image, mockup, logo, avatar, hero image, illustration, diagram, visual reference, or any generated picture, or when a design skill needs an image produced. Bridges to OpenAI's GPT Image 2.5 through the codex CLI using a ChatGPT subscription — no API key.
---

# gpt-image-bridge

This skill adds image generation to Claude Code by shelling out to the `codex` CLI and its `image_generation` tool. OpenAI's [Images 2.5 announcement](https://openai.com/index/introducing-chatgpt-images-2-5/) includes availability in Codex. Codex is authenticated via the user's ChatGPT subscription, so no API key is required.

The bridge uses the image model provided by Codex; it does not select or verify a particular backend model. The executable name `gpt-image-2` is retained for compatibility and does not pin the model to GPT Image 2. Do not pass an image-model ID to `codex --model`: that option selects the coding model, not the image backend.

## When to use

Invoke the wrapper any time the user wants an image produced. Examples:

- "Generate a hero image for..."
- "Make me a mockup of..."
- "I need an avatar / logo / illustration / diagram of..."
- "Show me what X would look like"
- A design-taste skill (e.g. `image-taste-frontend` from [Leonxlnx/taste-skill](https://github.com/Leonxlnx/taste-skill)) is driving an image-first workflow and needs an image generated

Do **not** invoke it for:

- Small programmatic PNGs (1px dots, solid-color placeholders, charts) — write those directly with a few lines of Python.
- Editing an existing image pixel-by-pixel.

## How to call it

```bash
~/.claude/skills/gpt-image-bridge/bin/gpt-image-2 "<prompt>" <absolute-output-path.png> [--size WxH]
```

- **Prompt** should be dense and art-directed: composition, lighting, camera/lens, mood, style reference. Terse prompts produce generic output.
- **Output path** must be absolute. `/tmp/` works for throwaways; a project-local `design/` directory for kept assets.
- **Size is optional** — if omitted, the model chooses its own dimensions. Only pass `--size` when the user specifies one or the layout requires a particular aspect ratio.
- Calls routinely take 4–6 minutes (codex reasons before calling the image tool; exact latency depends on the user's codex `reasoning_effort` config). Set the `Bash` tool timeout to the **maximum, 600000 ms**, or run it in the background and poll.
- The wrapper prints the absolute output path to stdout on success, or a tail of the codex log to stderr on failure.

After the wrapper returns, `Read` the PNG back into context so you can analyze it before coding or responding.

## Under the hood

The wrapper runs:

```

Report a problem