|
|
vor 1 Monat | |
|---|---|---|
| .. | ||
| coding_prompts | vor 1 Monat | |
| meta_prompts | vor 1 Monat | |
| system_prompts | vor 1 Monat | |
| user_prompts | vor 1 Monat | |
| .gitkeep | vor 1 Monat | |
| README.md | vor 1 Monat | |
i18n/zh/prompts/ stores the prompt assets of this repository: System Prompts are used to constrain AI's boundaries and taste, and Task Prompts drive the development pipeline of "demand clarification → planning → execution → review".
v8 or v10).coding_prompts/ by stage in specific tasks (clarification / planning / execution / review).skills/ (more reusable, more stable).i18n/en/prompts/
├── README.md
├── coding_prompts/ # Programming/R&D Prompts (currently 41 .md files)
│ ├── index.md # Auto-generated index and version matrix (do not modify manually)
│ ├── Standardized Process.md
│ ├── Project Context Document Generation.md
│ ├── Intelligent Requirement Understanding and R&D Navigation Engine.md
│ └── ...
├── system_prompts/ # System Prompts (Multiple versions of CLAUDE + other collections)
│ ├── CLAUDE.md/ # Version 1~10 directory (v9 currently placeholder only)
│ │ ├── 1/CLAUDE.md
│ │ ├── 2/CLAUDE.md
│ │ ├── ...
│ │ ├── 9/AGENTS.md # v9 currently no CLAUDE.md
│ │ └── 10/CLAUDE.md
│ └── ...
└── user_prompts/ # User's own/one-time prompts
├── ASCII Art Generation.md
├── Data Pipeline.md
└── Project Variables and Tools Unified Maintenance.md
system_prompts/: System-Level Prompts (First Make AI "Controllable")System prompts are used to define working mode, code taste, output format, and security boundaries. The directory adopts a versioned structure:
i18n/en/prompts/system_prompts/CLAUDE.md/<version_number>/CLAUDE.mdv8: Comprehensive version, suitable for general Vibe Codingv10: More focused on Augment/context engine standardization constraintsv9 directory is currently a placeholder only (no CLAUDE.md)coding_prompts/: Task-Level Prompts (Make the Process Run Through)coding_prompts/ are geared towards "one task": from requirement clarification, planning decomposition to delivery and review. It is recommended to treat it as a workflow script library:
Project Context Document Generation.md: Solidify context, reduce cross-session driftIntelligent Requirement Understanding and R&D Navigation Engine.md: Break down vague requirements into executable tasksStandardized Process.md: Hardcode "what to do first, what to do next", reduce loss of controlSystem Architecture Visualization Generation Mermaid.md: Output architecture as visualizations (a picture is worth a thousand words)index.md (Important)coding_prompts/index.md is an automatically generated index (including version matrix and jump links), do not edit manually. If you batch add/delete/adjust versions, it is recommended to generate the index through the toolchain and then synchronize.
user_prompts/: Personal Workbench (Not for Systematization)Place some personal habits, temporary scaffolding prompts here. The principle is: usable, not messy, not polluting the main library.
# View a task prompt
sed -n '1,160p' i18n/zh/prompts/coding_prompts/Standardized Process.md
# Select system prompt version (it is recommended to back up your current CLAUDE.md first)
cp i18n/zh/prompts/system_prompts/CLAUDE.md/10/CLAUDE.md ./CLAUDE.md
If you need batch maintenance capabilities for Excel ↔ Markdown, the repository has a built-in third-party tool: libs/external/prompts-library/. It is recommended to treat it as a "prompt asset production tool", and i18n/zh/prompts/ as a "curated collection for daily development".
../skills/: Consolidate high-frequency domain capabilities into Skills (stronger reuse)../documents/: Methodologies and best practices (prompt design and workflow principles)../libs/external/prompts-library/: Prompt Excel ↔ Markdown management tool