rosekhlifa/image-prompts

Image-Prompts is a bilingual (Chinese and English) website where AI image creators can discover, share, and reuse text prompts, the descriptive instructions that tell AI tools what to generate. Think of it as a community-driven library of inspiration: anyone can browse, search, favorite, and share prompts. Logged-in users can submit their own prompts, which go through an admin review process before being published. Each prompt can be sent directly to a companion desktop app called Image-Studio with a single click, so you can start generating images without manually copying and pasting text. The site pulls its content from three sources: user submissions, admin-curated archives, and publicly crawled datasets from other prompt galleries (with original source links preserved on every imported entry). As of the README, it has aggregated over 30,000 prompts from sources like Liblib, YouMind, and others. Content is organized by categories, tags, and sorting options (newest, popular, most liked, most sent to Studio). NSFW content is hidden by default and requires an explicit acknowledgment before viewing. The project includes a full admin dashboard for site management, handling user roles, bans, audit logs, site-wide configuration, content moderation via a report queue, and bulk JSONL imports up to 50 MB. It supports Google and GitHub login, user profiles with pinned works, and full Chinese/English translation throughout the interface. It's built as a monorepo with a React frontend and a Hono backend running on PostgreSQL, with image storage on Cloudflare R2 (credentials encrypted at rest). The README notes it doesn't require Docker and documents deployment on Ubuntu with 1Panel and OpenResty. The roadmap lists planned features including full-text search with bilingual tokenization, semantic similarity recommendations, comment threads, a public API, and a "remix tree" showing how prompts derive from each other. Who would use this? AI image creators who want a searchable, bilingual prompt reference, whether they're looking for inspiration, tracking down the source of a prompt they saw elsewhere, or building their own library of favorites. It's also aimed at site operators who want to run a similar prompt-aggregation community, since the codebase is open-source under MIT and includes deployment documentation.