A Curated Collection of GPT Store Prompts, Organized So You Don't Have To

If you've ever scrolled through the GPT Store looking for something genuinely useful, you know the feeling: endless listings, vague descriptions, and no easy way to peek under the hood at the prompts that actually make these things work. The GPTS-Prompt-Collection is a straightforward answer to that problem—a running archive of GPT Store prompts, sorted into categories, with the actual prompt text saved as readable markdown files.

What It Does

This repository collects prompts from GPT Store apps and stores them as plain markdown files in a GitHub repo. The project is organized around a category system, with separate folders for different use cases. According to the README, the collection is meant to be ongoing—the maintainer describes it as a continuous effort to gather interesting prompts from the GPT Store.

The categories listed in the README are:

• Writing
• Developer
• Productivity
• Business
• Education
• Finance
• Health
• Marketing
• Sales
• Games
• Designer

There's also a "Boutique" section for manually curated picks, which the README calls "人工精选" (hand-selected). That section includes prompts for things like a Dia browser system prompt, Manus agent and tools prompts, a Claude deep-thinking prompt, a study assistant, mind-map generation, translation, PDF-to-webpage conversion, algorithm competition problems, concept explanation, domain knowledge crash courses, and text summarization.

Each category folder holds individual .md files named after the GPT they came from. The Writing category alone has dozens of entries—everything from an "AI Novel" prompt to a "SEO Blog Content Writer" to a "Limerick" generator. The Marketing folder has entries like a "$100M Offer" prompt and a "Meta Adcopy Generator." The README also includes a star history chart, which suggests the project tracks its own growth over time.

Why It's Cool

The real value here isn't the prompts themselves—it's the organization. Anyone can copy a prompt from a GPT Store listing, but having them sorted by function and stored as plain text makes the collection actually usable.

You can read the prompts, not just use them. Most GPT Store apps are black boxes. You interact with them, but you don't see the instructions behind the curtain. This repo flips that. Every prompt is saved as a markdown file, which means you can study how these things are structured, what techniques they use, and how they handle edge cases. If you're trying to write your own prompts, this is a goldmine of examples.

The Boutique section is the highlight. The manually curated picks are more interesting than the bulk-collected ones. There's a Dia browser system prompt, a Manus agent prompt with separate files for tools and prompts, and a Claude deep-thinking prompt. These aren't just "write me a blog post" templates—they're system-level instructions for more complex tools. That's the kind of thing you don't usually get to see.

It's just markdown files in a repo. No app to install, no API keys, no account required. You can browse the collection directly on GitHub, clone it locally, or fork it and adapt the prompts for your own use. The simplicity is the point. There's no lock-in, no proprietary format, no "sign up to view the prompt" nonsense.

The category system is practical. Eleven categories might sound like a lot, but they map to real use cases. If you're working on a marketing campaign, you go to the Marketing folder. If you need a writing assistant, you go to Writing. The README also links directly to each folder, so you're not hunting through a file tree.

The collection is bilingual. Many of the prompts have Chinese names or descriptions, and some are in English. That's a nice bonus if you work across languages or want to see how prompt engineering is approached in different communities.

How to Try It

There's no installation or setup. The whole thing is a GitHub repository full of markdown files.

1. Go to the repository: https://github.com/b3o/gpts-prompt-collection

2. Browse the category folders in the README. The "Boutique" section is a good starting point if you want the hand-picked stuff.

3. Click into any .md file to read the prompt. You can copy the text directly from GitHub or clone the repo if you want to work with the files locally.

4. If you want to use a prompt, paste it into ChatGPT or whatever LLM you're working with. These are just text instructions—they're not tied to any specific platform.

5. If you find something useful or want to contribute, the README asks for a star to help track interest and updates.

There's also a sponsor section in the README linking to a shop for GPT Plus subscriptions and virtual credit cards, but that's separate from the prompt collection itself.

Final Thoughts

This isn't a polished product—it's a collection, and it shows. Some categories are deeper than others, and the quality varies from prompt to prompt. But that's also what makes it useful. It's a snapshot of how people are actually using GPT Store apps, organized in a way that lets you learn from it.

If you're into prompt engineering, this is a solid reference. If you're just looking for a specific prompt—say, a translation assistant or an SEO writer—you'll probably find something here. And if you're building your own GPTs, the Boutique section alone is worth a look. It's not flashy, but it's practical, and it's the kind of resource that gets better as more people contribute.

Follow @githubprojects for more developer tools and open source projects.