How yaojingang/yao-open-prompts works
1. Repository purpose
- This is a curated, documentation-first repository of AI prompts rather than an application with a runtime service.
- The primary artifact is a library of Markdown prompt files organized by use case and language.
2. Core content layer: prompts/
- prompts/ is the main Chinese prompt corpus.
- It is organized by numbered scenario groups such as AI methods, work, learning, content, marketing, coding, and thinking.
- Each prompt is a standalone Markdown document intended to be copied directly into an LLM workflow.
3. Parallel English mirror: prompts-en/
- prompts-en/ is an English-language mirror of the same taxonomy.
- The README states the English set follows the same paths and structure as prompts/, which makes the repository effectively bilingual and path-aligned.
- This mirror is important architecturally because it implies content generation/synchronization tooling rather than independent manual curation only.
4. Prompt file schema and content boundary
- Prompt files use a consistent frontmatter schema: title, category, subcategory, source_section, author, version, created, status, and tags.
- The body is intentionally constrained to the reusable prompt content, keeping tutorials, screenshots, and long explanations outside the prompt payload.
- This separation is the key boundary between “executable prompt text” and supporting documentation.
5. Taxonomy and navigation layer
- CATALOG.md is the full index over the repository and acts as the main navigation surface.
- Category counts in the README show the repository is organized around usage scenarios, not model families or technical domains.
- README.md and README.en.md are the top-level entry points; prompts/README.md and prompts-en/README.md are the language-specific hubs.
6. Derived documentation layer: docs/
- docs/ contains a static site entry point (index.html) and .nojekyll, indicating GitHub Pages-style static publishing.
- The docs site is a rendered navigation layer for browsing the prompt collection, not a separate application backend.
- It likely consumes the catalog and prompt structure as source material for a website-friendly view.
7. Automation and repository maintenance: scripts/
- scripts/check_repo.py is the quality gate for validating repository consistency.
- scripts/generate_catalog.py rebuilds the central index from the prompt files.
- scripts/generate_english_readmes.py suggests automated synchronization of English navigation/docs.
- scripts/generate_webpage.py produces the static web navigation artifact in docs/.
8. Source-to-derived data flow
- The source of truth is the Markdown prompt corpus plus its frontmatter metadata.
- Scripts transform that source into derived outputs: catalog, language readmes, and the web page.
- This means architecture is “content repository with generated indexes,” not “manually maintained documentation only.”
9. Reference and policy layer: references/ and maintenance/
- references/ stores supporting knowledge, external resources, and related concept notes that should not be mixed into prompt bodies.
- maintenance/ contains operational checklists for review and release discipline.
- These folders separate editorial governance from prompt content, which helps keep the main corpus clean and reusable.
10. Template layer: templates/
- templates/prompt-file-template.md defines the canonical structure for new prompt files.
- This is the structural contract for contributors and is what keeps the corpus uniform enough for automated indexing.
11. Editorial and release workflow
- The README describes a process of adding prompts, versioning them in place, updating CHANGELOG.md, regenerating indexes, and running repository checks.
- That establishes a controlled content lifecycle: authoring → validation → regeneration → publication.
- The release checklist in maintenance/ formalizes this as a repeatable publishing process.
12. Major logical subsystem boundaries
- Content corpus: prompts/ and prompts-en/
- Metadata/indexing: CATALOG.md, README files, frontmatter
- Derived publication: docs/
- Automation: scripts/
- Governance/supporting material: references/, maintenance/, templates/
13. Technologies and runtime implications
- The repository is primarily Markdown-based, with lightweight Python tooling for validation and generation.
- There is no evidence of a server-side application, database, or API runtime.
- The architectural center is therefore static content management and build-time generation, suitable for GitHub-based publishing.
14. How to diagram it
- Show the prompt Markdown files as the authoritative source.
- Show scripts consuming those files to generate catalog/readme/web docs.
- Show references/ and maintenance/ as supporting governance inputs, not part of the prompt execution path.
- Show the published static site in docs/ as the user-facing browsing surface, while the prompt files remain the copyable operational payload.
Main entry points: prompts Chinese prompts · prompts-en English prompts · check_repo.py Repo check · prompt-file-template.md Prompt schema
Components 18
Prompt corpus
Authoritative Markdown prompt collections in Chinese and English.
- Chinese prompts · Authoritative Chinese prompt files organized by scenario taxonomy.
prompts - English prompts · English-language mirror aligned to the Chinese prompt paths and structure.
prompts-en - Prompt schema · Shared frontmatter fields that make prompts indexable and consistent.
templates/prompt-file-template.md
Navigation and publishing
Top-level indexes and the static browsing surface.
- README · Primary Chinese repository overview and contributor guidance.
README.md - README EN · English repository overview and mirrored navigation entry.
README.en.md - Catalog · Full repository index over the prompt collection.
CATALOG.md - Prompts hub · Entry point for the Chinese prompt taxonomy.
prompts/README.md - English hub · Entry point for the English prompt taxonomy mirror.
prompts-en/README.md - Docs site · GitHub Pages-style static navigation site.
docs/index.html
Build scripts
Validation and generation utilities that derive indexes and published artifacts.
- Repo check · Validates repository consistency and content structure.
scripts/check_repo.py - Catalog builder · Regenerates the central catalog from prompt files.
scripts/generate_catalog.py - EN sync · Generates or synchronizes English readme navigation files.
scripts/generate_english_readmes.py - Web builder · Produces the static web navigation artifact in docs/
scripts/generate_webpage.py
Support material
Templates, references, and operational governance documents.
- Prompt template · Canonical structure for new prompt files.
templates/prompt-file-template.md - References · Background notes and external resources kept separate from prompts.
references - Maintenance · Review and release checklists for prompt curation.
maintenance - Changelog · History of prompt updates and repository releases.
CHANGELOG.md - Contributing · Contributor guidance for adding and maintaining prompts.
CONTRIBUTING.md
Connections 23
- README → Chinese prompts: introduces · Chinese README points readers to the main corpus.
- README EN → English prompts: introduces · English README points readers to the mirrored corpus.
- Prompts hub → Chinese prompts: navigates · Language hub organizes the Chinese prompt collection.
- English hub → English prompts: navigates · Language hub organizes the English mirrored collection.
- Prompt schema → Chinese prompts: defines · Template establishes the common metadata contract for prompt files.
- Prompt schema → English prompts: defines · Template also constrains the mirrored English prompt files.
- Chinese prompts → Catalog: indexed by · Prompt corpus is the source for the global catalog.
- English prompts → Catalog: indexed by · Mirrored corpus also participates in the repository index.
- Catalog builder → Catalog: builds · Catalog generator rebuilds the central index.
- EN sync → English hub: generates · English sync script produces language navigation files.
- Web builder → Docs site: builds · Web generator produces the published static browsing surface.
- Repo check → Chinese prompts: validates · Repository check enforces corpus consistency.
- Repo check → English prompts: validates · Repository check also covers the mirrored corpus.
- Repo check → Prompt schema: checks · Validation depends on the expected prompt schema.
- Catalog builder → Chinese prompts: reads · Catalog generation consumes the Chinese source files.
- Catalog builder → English prompts: reads · Catalog generation consumes the English mirror as well.
- EN sync → Chinese prompts: mirrors · English docs are synchronized from the source taxonomy.
- Web builder → Catalog: consumes · Static site generation uses the catalog as navigation source.
- References → Chinese prompts: supports · Reference material informs prompt writing without becoming prompt payload.
- Maintenance → Changelog: governs · Release process links review workflow to version history.
- Maintenance → Repo check: requires · Operational checklists rely on validation before release.
- Prompt template → Prompt schema: standardizes · Template encodes the canonical file structure.
- Contributing → Prompt template: guides · Contributor instructions point to the canonical prompt template.