The GPTS Prompt Collection: A Treasure Trove for OpenAI's Custom GPT Era
You've probably spent hours tinkering with custom GPTs in the ChatGPT interface, tweaking system prompts and wondering if you're missing some clever trick that would make your bot genuinely useful. The problem isn't that you can't write prompts—it's that you don't know what's possible. That's where the GPTS-Prompt-Collection comes in. It's a curated, categorized repository of real prompts harvested from the GPTS Store, and it's exactly the kind of reference you wish you had when you started.
What It Does
This repository is exactly what its name suggests: a collection of prompts from OpenAI's GPTS Store, organized by category and continuously updated. The project's README (written in Chinese, though the content is largely language-agnostic) describes it as an ongoing effort to collect "wonderful prompts" from the store, with a request for stars to keep the project visible.
The structure is straightforward. The repository is divided into folders by use case: Writing, Developer, Productivity, Business, Education, Finance, Health, Marketing, Sales, Games, and Designer. There's also a "Boutique" folder for hand-picked, manually collected prompts—the curator's personal favorites. Within each folder, you'll find individual markdown files, one per prompt, with names like "AI作文助手" (AI Writing Assistant), "SEO博客内容编写器" (SEO Blog Content Writer), or "TEDTalk Writer."
The project also includes a few notable extras. The Boutique folder contains what appear to be system prompts for other AI agents like Manus and Dia browser, plus some general-purpose prompts for deep thinking, study assistance, mind mapping, and translation. There's a star history chart embedded in the README, which is a nice touch for gauging community traction. And yes, there's a sponsorship section with affiliate-style offers for GPT Plus and virtual credit cards—a common way open-source projects monetize, though you can ignore it if that's not your thing.
Why It's Cool
The value here isn't in any single prompt—it's in the aggregate. Let me break down why this repository is worth your attention.
It's a learning resource disguised as a collection. Reading other people's system prompts is one of the fastest ways to understand how to structure your own. You'll see how different creators handle constraints, personality, output formatting, and guardrails. It's like reading well-commented code written by other developers.
The categorization is thoughtful. The folder structure maps directly to practical use cases. If you're building a marketing GPT, you can jump straight into the Marketing folder and find prompts like "LinkedIn广告专家" (LinkedIn Ad Expert) or "Meta Adcopy Generator." The naming is mostly in Chinese, but the content is a mix—you'll find prompts in English, Chinese, and bilingual formats.
The Boutique folder is genuinely curated. Unlike the category folders which seem to be bulk-collected, the Boutique section contains hand-picked prompts covering things like agent system prompts for Manus, browser system prompts, deep thinking frameworks, and study aids. That's the kind of high-signal content that makes browsing worthwhile.
It's a snapshot of the GPT ecosystem. The GPTS Store is a fascinating experiment in prompt engineering as a product. This collection gives you a window into what's actually popular and useful in that space—not just what OpenAI markets, but what real creators build and share.
It's actively maintained. The README emphasizes continuous updates, and the star history chart suggests a project with real community interest. That means the collection is likely to grow and stay relevant.
How to Try It
Getting started is as simple as cloning the repository or browsing it directly on GitHub.
git clone https://github.com/B3o/GPTS-Prompt-Collection.git
cd GPTS-Prompt-Collection
From there, you can browse the folders to find prompts that interest you. Each markdown file contains the full prompt text, which you can copy directly into your custom GPT instructions or use as inspiration for your own prompts.
If you're building a custom GPT, here's a practical workflow:
- Identify the category that matches your use case (e.g.,
./14/for Writing). - Open a few prompt files that sound relevant to what you're building.
- Study the structure—how the prompt sets context, defines constraints, and structures output.
- Adapt the best elements into your own system prompt.
You can also check the Boutique folder for general-purpose prompts that aren't tied to a specific category. The "深度思考" (deep thinking) and "学习助手" (study assistant) prompts are particularly useful as starting points for more open-ended GPTs.
If you find the collection useful, consider starring the repository—it helps with visibility and encourages the maintainer to keep updating.
Final Thoughts
This is a niche resource, but for anyone working with custom GPTs, it's a genuinely useful one. The prompts range from practical to quirky (there's a "你的前女友丽莎" or "Your Ex-Girlfriend Lisa" in the Writing folder, which is both hilarious and a good example of persona-driven prompts). The quality varies, as you'd expect from a crowd-sourced collection, but that's part of the value—you get to see what works and what doesn't.
This project is best for developers and creators who want to level up their prompt engineering by studying real examples. It's not a polished documentation site or a framework—it's a reference library. And sometimes, that's exactly what you need.
The GPTS ecosystem is still young, and collections like this are charting its territory as it grows. Whether you're building your first custom GPT or refining your tenth, you'll likely find something here worth stealing.
Follow @githubprojects for more developer tools and open source projects.