phoenix-gh

Prompix — Development skill for Claude Code

Development community

970+ image-generation prompts across 6 models — same dataset served as a web gallery, an MCP server, and a Claude Code / Codex skill.

How to install Prompix

This entry records only its repository, not the path inside it, so there is no exact command to give. Open phoenix-gh/prompix and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Prompix does

970+ image-generation prompts across 6 models — same dataset served as a web gallery, an MCP server, and a Claude Code / Codex skill. Bilingual (en/zh), attribution preserved, MIT + CC BY 4.0.

Alternatives in Development

Third-party Notices — THE FOLLOWING SETS FORTH ATTRIBUTION NOTICES FOR THIRD PARTY SOFTWARE THAT MAY BE CONTAINED IN PORTIONS OF THI 94.7k ★
Ars Abstract — ARS academic-paper abstract-only mode — bilingual abstract + keywords 4.5k ★
Petdex — A public gallery of animated pets for Codex, Claude Code, DeepSeek Harness, Hermes, OpenCode, Gemini CLI, and 4k ★

README

prompix

[![License (code): MIT](https://img.shields.io/badge/License%20%28code%29-MIT-blue.svg)](LICENSE) [![License (data): CC BY 4.0](https://img.shields.io/badge/License%20%28data%29-CC%20BY%204.0-lightgrey.svg)](LICENSE-DATA) [![Prompts: 970+](https://img.shields.io/badge/Prompts-970%2B-gold.svg)](data/prompts.json) [![Models: 6](https://img.shields.io/badge/Models-6-purple.svg)](data/prompts.json)

A curated, multi-model image-generation prompt library — with a web gallery, an MCP server, and an agent skill — so that humans **and** AI agents (Claude Code, Codex, Cursor, etc.) can discover and reuse great prompts.

Maintained by **MK@Phoenix** · [@FongNb](https://x.com/FongNb) on X

Overview

prompix unifies **970+ prompts** from five upstream community projects (see [Acknowledgments](#acknowledgments)) into a single canonical schema:

One source of truth — data/prompts.json with a 13-category taxonomy, flat tag namespace, and per-prompt model labels covering GPT-Image-2, GPT-Image-1, GPT-4o, Nano Banana, Nano Banana Pro, and model-agnostic prompts.
Three surfaces, same data — a React gallery for humans, an MCP server (stdio + Streamable HTTP) for any agent, and an installable Claude / Codex skill for agents that prefer skills over MCP.
Bilingual end-to-end — titles, descriptions, guidance, and pitfalls carry both en and zh where the upstream provides them; the MCP tools accept a lang parameter.
Attribution preserved, license clean — every prompt keeps its upstream source.ref and author credit. Code is MIT; the dataset is CC BY 4.0. See `NOTICE.md`.