Yao Open Prompts: a bilingual prompt library with a Python check pipeline
Practical bilingual AI prompt library for prompt engineering, work, learning, content, marketing, and everyday use.
At a glance
- What is it?
- Yao Open Prompts is a CC BY 4.0 collection of 118 Chinese prompt files with a mirrored English set, organised by scenario and validated by two Python scripts. It suits prompt engineers who want copyable prompt bodies, not tutorial pages.
- Who is it for?
- Adopt Yao Open Prompts if you want a scenario-indexed, bilingual prompt corpus whose files you can copy into your own tooling, and if you are willing to run python3 scripts/check_repo.py after editing.
- Can I use it commercially?
- Check first. The repository uses a licence we do not classify automatically, so read its LICENSE file before any commercial use.
- Is it still maintained?
- Yes. The repository last received commits 49 days ago.
- What is it written in?
- Mainly Python, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on September 15, 2026, and from our analysis. They are not legal advice.
DEEP OPEN-SOURCE ANALYSIS
What problem the 118-file prompt library solves
Prompt collections usually arrive as one long article. You scroll, you copy a block, you lose the surrounding context, and the next time you need it you scroll again. Yao Open Prompts is the open-source version of 《姚金刚提示词合集》, reorganised so that each prompt lives in its own Markdown file under a scenario directory. The README states the repository holds 118 Chinese prompt files, reclassified by scenario, with the copyable body kept and the promotional material, result screenshots, video attachment notes and leftover HTML styling removed.
The intended reader is someone doing prompt engineering, content operations, learning design or marketing who wants a starting point rather than a blank page. The categories in the README run from AI方法 (8 files) and AI工作 (10) through AI学习 (11), AI生活 (2), AI教育 (4), AI内容 (50), AI编程 (1), AI营销 (29) and AI思考 (3). That distribution is the honest signal about the project: it is weighted heavily toward content and marketing, and barely present in programming. If your work is code generation, one file in the AI编程 category is not a library.
How prompts/, prompts-en/ and the Python scripts fit together
The data flow is file-based and deliberately flat. prompts/ holds the Chinese prompt bodies by category, and prompts-en/ mirrors the same paths with the English versions; the README says the English set is a complete mirror of prompts/ under the same relative paths. CATALOG.md is the full index generated from those files, and the website at yaojingang.github.io/yao-open-prompts/ is generated too, not hand-written.
Each prompt file carries frontmatter with a fixed key set, and the README says the body keeps only three parts: title, introduction and Prompt. That constraint is the design decision worth noting. Case studies, evaluation screenshots, tutorial links and long explanations are pushed to references/ or a future case directory instead of sitting next to the text you copy. The trade-off is that a prompt file tells you nothing about how well it performed; you get the prompt and no evidence.
The scripts close the loop. scripts/check_repo.py performs the quality check, scripts/generate_catalog.py rebuilds the index, and scripts/generate_webpage.py rebuilds the site. The README does not document what check_repo.py verifies beyond calling it a quality check, so treat its exact rules as something to read in the script itself.
Installing and running your first check
There is no package to install. The README gives no pip, npm or Docker step; the workflow is clone the repository, copy a prompt, and run the Python scripts that live in scripts/ when you change something. The commands below are the ones the README lists under 持续更新机制.
Clone the repository and confirm the three entry points exist before you change anything.
git clone https://github.com/yaojingang/yao-open-prompts.git
cd yao-open-prompts
ls scripts/You should see check_repo.py, generate_catalog.py and generate_webpage.py in the listing. The README does not state a minimum Python version, so if the scripts fail on an old interpreter, that is where to look first.
Run the quality check after editing any prompt file.
python3 scripts/check_repo.pyThe README describes this as the quality check step and does not publish its exit codes or error format, so read the output rather than assuming a silent pass means everything is valid.
When you add a prompt, the README says to copy the template into the matching category directory, then regenerate the index.
cp templates/prompt-file-template.md prompts/06-ai-content/my-new-prompt.md
python3 scripts/generate_catalog.pyThe template carries the frontmatter keys the repository expects: title, category, subcategory, source_section, author, version, created, status and tags. Fill them before regenerating, because the catalog is built from the files.
Where this library breaks down
The README is explicit that third-party or reposted content is not merged into the main prompt library. It goes to references/ first or gets marked third-party-review, and is published only after authorisation is confirmed. That is a defensible policy, and it also means the main prompts/ tree is not a complete record of prompt engineering practice; it is one author's curated set plus whatever cleared review.
Versioning is per file, not per release. The README's update rule is to raise version inside the original file and record the change in CHANGELOG.md, and it recommends date-based releases such as v2026.05.1. There is no retrieved release in the repository data, so there is no tagged artifact to pin against. If you need to reproduce a prompt exactly as it was six months ago, you depend on the file history, not on a release page.
The file-per-prompt structure has a second cost. Series content is merged into thematic collections on purpose, with the README citing 50个 Nano Banana 创意提示词 and the Nano Banana PPT collection as examples, to stop the directory from fragmenting into many short files. Merging keeps the tree readable, but it means one file can hold dozens of prompts, so a single frontmatter version covers a batch that may change at different rates.
And the library is not a runtime. Nothing here calls a model. If you want prompt execution, evaluation harnesses or a serving layer, this repository gives you the text and the directory conventions, and stops there.
Yao Open Prompts versus a prompt framework such as Geoflow
The closest thing to a comparison in the search data around this project is Geoflow, which sits in a different category. Geoflow is a prompt framework: prompts are inputs to an execution pipeline, and the value is in running them. Yao Open Prompts is a content repository. The README's own structure makes the boundary clear, with prompts/, prompts-en/, references/, templates/, maintenance/, scripts/ and docs/ as top-level directories, and no runtime, service or API described anywhere in it.
The practical difference shows up when you want to change something. In a framework you edit a node in a pipeline and rerun it. Here you edit a Markdown file, raise its version key, add a CHANGELOG.md entry, then run python3 scripts/check_repo.py and python3 scripts/generate_catalog.py so the index and the generated site stay consistent with the files. That is a publishing workflow, not an execution workflow. If your need is orchestration, a repository of prompt bodies will not substitute for it, and if your need is a well-indexed set of prompts you can read and copy, a framework adds machinery you would have to maintain.
Maintenance cost, licence and what the repository does not settle
The last push to the default branch was on 2026-07-29, and the repository is not archived. The README describes an ongoing mechanism rather than a finished state: new prompts come from templates/prompt-file-template.md, updates raise the version field and land in CHANGELOG.md, category changes require editing category, subcategory and tags in the frontmatter and regenerating the catalog, and releases are meant to follow maintenance/release-checklist.md with date-based tags. The recent additions listed in the README span generators, meta-prompts, GEO, learning, products and a marketing collection, which is a broad spread for a single maintainer to keep at a consistent quality bar.
On licensing, the repository metadata reports NOASSERTION, while the README states that prompt content uses CC BY 4.0 and that any future scripts or tool code may use MIT separately. Those two statements do not match, so read the LICENSE file in the repository root and the frontmatter of the specific file you intend to reuse before you assume either. This is a description of what the repository says, not legal advice.
The FAQ below covers the questions a new user is most likely to bring to this project.
Editorial conclusion
Adopt Yao Open Prompts if you want a scenario-indexed, bilingual prompt corpus whose files you can copy into your own tooling, and if you are willing to run python3 scripts/check_repo.py after editing. Do not adopt it if you need an executable framework, an API, or prompts with per-file provenance for third-party content: the README states that obvious third-party or reposted material is kept in references/ or marked third-party-review rather than merged into the main library. Before relying on it, verify the LICENSE file, since the repository metadata reports NOASSERTION while the README states CC BY 4.0 for prompt content and mentions MIT for any future script or tool code.
Frequently asked questions
How do I install Yao Open Prompts?
There is no install step. The README gives no package or container command; you clone the repository and copy the Prompt section out of the Markdown file you need, then run the Python scripts in scripts/ only if you are editing the repository itself.
How do I use a prompt from Yao Open Prompts?
The README lists four steps: find the prompt by scenario in CATALOG.md, open the Markdown file and copy the Prompt area, replace the variables or placeholders or example content with your real task information, then test the output in your target model and record version iterations.
Is there an English version of the Yao Open Prompts library?
Yes. README.en.md is the English documentation entry point and it navigates to 118 English prompts in prompts-en/, which the README says mirror the same paths as prompts/.
Community notes