[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-79902":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":8,"htmlUrl":8,"language":9,"languages":8,"totalLinesOfCode":8,"stars":10,"forks":11,"watchers":12,"openIssues":13,"contributorsCount":13,"subscribersCount":13,"size":13,"stars1d":13,"stars7d":13,"stars30d":14,"stars90d":13,"forks30d":13,"starsTrendScore":13,"compositeScore":15,"rankGlobal":8,"rankLanguage":8,"license":8,"archived":16,"fork":16,"defaultBranch":17,"hasWiki":18,"hasPages":16,"topics":19,"createdAt":8,"pushedAt":8,"updatedAt":20,"readmeContent":21,"aiSummary":22,"trendingCount":13,"starSnapshotCount":13,"syncStatus":23,"lastSyncTime":24,"discoverSource":25},79902,"gpt-image-2-pro-max","therichardngai-code\u002Fgpt-image-2-pro-max","therichardngai-code",null,"Python",96,35,92,0,4,4.67,false,"main",true,[],"2026-06-12 02:03:55","# gpt-image-2-pro-max — Claude Code skill\n\nProduction prompt-engineering skill for **GPT-Image-2 \u002F OpenAI image generation**. Pairs a `media-designer` agent with a hosted searchable corpus of 7,405 community-vetted prompts (9,903 reference images).\n\nWe collect high-quality prompts and image examples for GPT-Image-2 across portraits, posters, character sheets, UI mockups, and community experiments.\n\nMost cases in this repository are curated from X\u002FTwitter, creator communities, public demos, and shared experiments.\n\nIf you find this useful, consider giving it a star. ⭐\n\n## What you get\n\n- **Agent**: `.claude\u002Fagents\u002Fmedia-designer.md` — diagnoses your brief, picks a mood-aligned base, refactors it into a parameterised template, and returns a paste-ready prompt.\n- **Skill + tool**: `.claude\u002Fskills\u002Fgpt-image-2-pro-max\u002F` — `search.py` thin client over a hosted backend (BM25 search across 10 facet vocabularies).\n- **Companion skill**: `.claude\u002Fskills\u002Fmedia-tools\u002F` — optional CLI for actually generating images (multi-provider chain: OpenRouter → Gemini → OpenAI gpt-image-2 → MiniMax → DashScope → BytePlus, or your ChatGPT Plus session). See its own `README.md` for setup.\n- **Hosted backend** (already live): `https:\u002F\u002Fgpt-image-2-prompts.goclawoffice.com`\n  - 7,405 indexed prompts and 9,903 reference images with BM25 search\n  - Rate-limited per IP for fair use; a small set of guards keeps the corpus available to everyone (origin\u002FUA gate, ≥3-token queries, daily budgets, edge cache)\n\n## Install\n\nDrop the `.claude\u002F` folder into your Claude Code project. No build step, no API key, no install.\n\n```bash\n# Clone or copy\ncp -r .claude \u002Fpath\u002Fto\u002Fyour-project\u002F\n\n# Verify the skill is discoverable\nls \u002Fpath\u002Fto\u002Fyour-project\u002F.claude\u002Fskills\u002Fgpt-image-2-pro-max\u002F\n```\n\nPython 3.8+ required (uses stdlib only — no pip install).\n\n## Usage\n\nIn Claude Code, invoke the skill directly or let the `media-designer` agent drive:\n\n```bash\n# Free-text search (BM25 + tag-aware ranking; needs ≥3 descriptive tokens)\npython .claude\u002Fskills\u002Fgpt-image-2-pro-max\u002Fscripts\u002Fsearch.py \"luxury shoe ad cream pastel\" -n 5\n\n# Narrow by output shape, require a reference image\npython .claude\u002Fskills\u002Fgpt-image-2-pro-max\u002Fscripts\u002Fsearch.py \"moody cinematic portrait 35mm\" \\\n    --shape portrait --has-image -n 3\n\n# Show full prompt body for every rank (top-1 is always full)\npython .claude\u002Fskills\u002Fgpt-image-2-pro-max\u002Fscripts\u002Fsearch.py \"neon ui dashboard\" --full -n 3\n\n# Persist top hits as a markdown reference deck (with embedded images)\npython .claude\u002Fskills\u002Fgpt-image-2-pro-max\u002Fscripts\u002Fsearch.py \"anime portrait\" --persist refs.md\n\n# Browse the controlled vocabulary (10 facets: subjects, styles, lighting,\n# cameras, moods, palettes, compositions, mediums, techniques, usecases)\npython .claude\u002Fskills\u002Fgpt-image-2-pro-max\u002Fscripts\u002Fsearch.py --list facets\npython .claude\u002Fskills\u002Fgpt-image-2-pro-max\u002Fscripts\u002Fsearch.py --list moods\npython .claude\u002Fskills\u002Fgpt-image-2-pro-max\u002Fscripts\u002Fsearch.py --list palettes\n```\n\nThe agent profile is canonical — its `diagnose → search → pick → refactor → resolve` loop is how good prompts get built.\n\n## Attribution\n\nEach record carries its original tweet\u002FX link in the `tweet` field. The agent profile makes attribution mandatory in every output. Prompt content belongs to the original authors.\n\n### Credits\n\nBig thanks to the **[EvoLinkAI\u002Fawesome-gpt-image-2-prompts](https:\u002F\u002Fgithub.com\u002FEvoLinkAI\u002Fawesome-gpt-image-2-prompts)** community for curating and openly sharing a large portion of the prompt corpus that seeds this skill. Their work is what makes this useful out of the box.\n\nIf you use this skill, please also star their repo — most prompts here trace back to authors they collected and credited first.\n\nThe `media-tools` companion skill's **ChatGPT \u002F Codex OAuth image-generation** path is a Python port of the pattern landed in **[nextlevelbuilder\u002Fgoclaw#1002](https:\u002F\u002Fgithub.com\u002Fnextlevelbuilder\u002Fgoclaw\u002Fpull\u002F1002)** — same priority-chain idea (skip provider if no key → first success wins → cascade on failure), same OAuth-token reuse trick so users with an existing ChatGPT Plus subscription don't need a separate API key. Credit and gratitude to the goclaw maintainers.\n\n## License\n\nPrompt content belongs to the original authors (Twitter\u002FX handles in each record). Skill code is MIT.\n","gpt-image-2-pro-max 是一个用于GPT-Image-2\u002FOpenAI图像生成的生产级提示工程工具。它提供了一个媒体设计师代理，可以诊断用户需求、选择符合情绪基调的基础模板，并将其重构为参数化模板以生成可直接使用的提示。此外，该项目还配备了一个基于BM25搜索算法的后端服务，支持对7,405个社区验证过的提示（9,903张参考图片）进行高效检索。此项目特别适用于需要高质量图像生成场景，如肖像画、海报设计、角色表单创建及UI原型制作等。其安装简便，仅需将`.claude\u002F`文件夹复制到你的Claude Code项目中即可使用，无需额外构建步骤或API密钥。",2,"2026-06-11 03:58:26","CREATED_QUERY"]