songtianlun/awesome-prompts

What it solves

Writing good prompts for AI models is tricky — especially when you need to generate images or videos, not just text. This project gathers verified, well-tested prompts in one place, organized by what you want to create (text, image, or video) and by task type, so you don’t have to start from scratch or guess what works.

How it works

The collection is split into two parts: the main library on the website prmbr.com and a visual gallery at prmbr.com/cases/. The gallery stores each example as a structured record in a JSON file (docs/data/cases.json), with fields like title, category, model, tags, the original prompt, a Chinese translation, usage notes, and a source URL. Preview images are stored as WebP files in docs/images/cases/. Older Markdown-based collections are still linked from the gallery as legacy collections. The latest batch adds 55 verified cases: 27 from Google's Nano Banana image documentation, 24 from OpenAI's GPT Image prompting guide, and 4 from the Sora 2 prompting guide. Cases that couldn't be redistributed (e.g., blog images without licenses) or where a prompt couldn't be tied to a specific result were left out.

Who it’s for

Anyone who works with AI models — developers, designers, creators, or hobbyists — who wants ready-to-use, proven prompts for text, image, or video generation. It’s also useful for people who want to learn prompt writing by example, especially since each case includes usage notes and a translation. The project also offers community support via WeChat and QQ groups.

Highlights

• Organized by modality (text, image, video) and task type
• 55 newly added, verified cases from official sources (Google, OpenAI)
• Each case includes the prompt, a Chinese translation, usage notes, and a source link
• Visual gallery with WebP previews for image/video cases
• Structured JSON schema for easy browsing or extending
• Active community with WeChat and QQ groups