# gpt-image-cookbook

> A multi-provider AI image generation cookbook bundling a curated prompt gallery, an agentic skill (`SKILL.md`), and a Python CLI (`gic`) into one repository. Supports OpenAI gpt-image-2, Google Imagen, and Flux (fal.ai / Replicate) under a unified interface for text-to-image, reference-image edits, inpainting, and multi-reference workflows.

## Repository

- **Name**: gpt-image-cookbook
- **Owner**: Eugeniu Ghelbur ([@eugeniughelbur](https://github.com/eugeniughelbur))
- **License**: MIT
- **Language**: Python (≥3.11)
- **CLI binary**: `gic`
- **Repository URL**: https://github.com/eugeniughelbur/gpt-image-cookbook

## What this provides

1. **Prompt gallery** — copy-paste prompts organized by category (posters, UI mockups, photography, diagrams, illustration, brand systems, edit/inpaint workflows). Each entry documents provider, model, quality, size, and the exact prompt.
2. **Agent skill** — `skills/gpt-image-cookbook/SKILL.md` is a runbook for Claude Code, Codex, OpenClaw, Hermes, and other skill-capable agent runtimes. It tells the agent how to classify the request, search the gallery, refine the prompt, and call `gic`.
3. **CLI (`gic`)** — one command, multi-provider. Switch providers with `--provider {openai,imagen,flux}`. Reads keys from process env, then `./.env`, then `~/.env`.

## Provider support matrix

| Provider | Models | Status |
|---|---|---|
| OpenAI | gpt-image-2 | implemented (text-to-image, edits, inpaint, multi-reference) |
| Google | imagen-4, imagen-4-fast | stub — interface defined, implementation pending |
| Flux | flux-pro-1.1, flux-schnell | stub — interface defined, implementation pending |

## Installation

```bash
pip install gpt-image-cookbook        # PyPI (planned)
uvx --from git+https://github.com/eugeniughelbur/gpt-image-cookbook gic --help   # one-shot
```

## Claude Code plugin

Installable via `/plugin install eugeniughelbur/gpt-image-cookbook`. The plugin loads `skills/gpt-image-cookbook/SKILL.md` and resolves `gic` automatically.

## Key files for LLMs and agents

- [README.md](https://github.com/eugeniughelbur/gpt-image-cookbook/blob/main/README.md): user-facing docs
- [skills/gpt-image-cookbook/SKILL.md](https://github.com/eugeniughelbur/gpt-image-cookbook/blob/main/skills/gpt-image-cookbook/SKILL.md): agent runbook
- [skills/gpt-image-cookbook/references/gallery.md](https://github.com/eugeniughelbur/gpt-image-cookbook/blob/main/skills/gpt-image-cookbook/references/gallery.md): gallery routing index
- [skills/gpt-image-cookbook/references/craft.md](https://github.com/eugeniughelbur/gpt-image-cookbook/blob/main/skills/gpt-image-cookbook/references/craft.md): prompt-craft cheatsheet
- [skills/gpt-image-cookbook/references/providers.md](https://github.com/eugeniughelbur/gpt-image-cookbook/blob/main/skills/gpt-image-cookbook/references/providers.md): per-provider semantics + capability matrix
- [src/gic/cli.py](https://github.com/eugeniughelbur/gpt-image-cookbook/blob/main/src/gic/cli.py): CLI entry point
- [src/gic/providers/](https://github.com/eugeniughelbur/gpt-image-cookbook/blob/main/src/gic/providers/): provider abstraction + implementations
