gpt-2-image-skill

OpenAI GPT Image 2 skill — curated prompt templates + CLI. Set OPENAI_API_KEY in env (or ~/.env) before use.

Install

Claude Code (plugin):

/plugin marketplace add wuyoscar/gpt_image_2_skill
/plugin install gpt-image@wuyoscar-skills

Codex / other SKILL.md-aware agent — clone into the agent's skills directory:

git clone https://github.com/wuyoscar/gpt_image_2_skill.git

Point the agent at skills/gpt-image/SKILL.md inside the clone.

Build your own custom skill on top — use the skill-creator skill:

/skill-creator
> Please help me create a skill based on https://github.com/wuyoscar/gpt_image_2_skill

CLI anywhere (no agent needed):

uvx --from git+https://github.com/wuyoscar/gpt_image_2_skill gpt-image -p "a cat astronaut"

# or install to PATH
uv tool install git+https://github.com/wuyoscar/gpt_image_2_skill
gpt-image -p "a cat astronaut"

Requires uv + Python ≥ 3.11. Deps auto-install.

Showcase

All images produced one-shot at --quality high. Every prompt is in references/gallery.md; captions below credit the original source so you can trace it.

Chinese typography & posters

Chinese tea-launch poster · community (卡尔的AI沃茨)	1980s propaganda poster · community (@akokoi1) + adapted
Museum-catalog infographic (唐代襦裙) · community (@MrLarus)	Song Dynasty social-media feed (宋朝朋友圈) · community (@Panda20230902)

Editorial & design posters

Saul-Bass-style thriller poster · original (Bass lineage)	Manga relationship map — A Tale of Two Cities · community (@cht0001)
Vogue-style fashion magazine cover · original	1950s sci-fi pulp cover (Astounding Stories) · original

Anime & manga

MAPPA-style anime action still · aesthetic ref: Jujutsu Kaisen	Shōnen battle key-visual · aesthetic ref: Naruto Shippuden
Shōnen manga two-page spread (basketball dunk) · original, Slam-Dunk tradition

Infographics & field guides

Encyclopedia field guide (Giant Panda) · community (@MrLarus)

Product & food photography

Chocolate wafer product render (JSON-style) · community (@mehvishs25)

Salad-explosion food photo (JSON-style) · community (@ChillaiKalan__)

Gaming landscapes

Hitman gameplay — OpenAI HQ · community (@flowersslop)

GTA 6 gameplay — Vice City beach · community (@WolfRiccardo)

Photorealism & reportage

RAW iPhone — 42nd Street subway · community (@WolfRiccardo)	Handwritten notebook flatlay · community (@patrickassale)
Chess board mid-game · community (@EddGorenstein)	360° equirectangular panorama · community (@AIimagined)

Cinematic & animation

Pixar-style 3D still (kitten + soufflé) · community style pattern	1940s film-noir still · original (Alton / Toland lineage)
Professional 6-panel storyboard (Tokyo rooftop chase) · original

Character design

Cyberpunk engineer reference sheet · community (@MANISH1027512)

16-panel anime expression grid · community (卡尔的AI沃茨)

Research paper figures

Transformer encoder–decoder · Vaswani et al., 2017	Retrieval-Augmented Generation pipeline · Lewis et al., 2020
Multi-agent LLM system · AutoGen (Wu et al., 2023) / LangGraph / Managed Agents	Denoising diffusion forward / reverse chain · Ho et al., 2020
Empirical scaling laws · Kaplan 2020, Chinchilla (Hoffmann et al., 2022)	Benchmark comparison heatmap · HELM (Liang et al., 2023)
Ablation bar chart · generic reasoning-paper layout	LLM pretraining data-mixture sankey · GPT-3 / LLaMA / Pile compositions
Multi-head attention heatmaps · Clark et al., 2019	Frontier LLM lineage, 2018–2026 · community survey format
ReAct reasoning trace · Yao et al., 2022

Acknowledgements

• OpenAI — the gpt-image-2 model.
• ZeroLu/awesome-gpt-image — community prompt archive, CC BY 4.0.
• ResearAI/AutoFigure — related LLM scientific-figure project.

Per-image community @handle credits and per-prompt research-paper citations live inline in references/gallery.md and in each Showcase caption above — no need to duplicate them here.

License

CC BY 4.0 — see LICENSE.