🛍️ 更多精选好物 ›
💻 Shell · MIT · 🐛 开放 Issue：2
📝 工具简介
精选的实用 AI 提示词合集，适用于各种任务和场景。
📖 怎么用
这是提示词库，无需安装。直接浏览 `useful-ai-prompts/prompts/` 目录按分类找提示词，或用 Python 读取索引：`python -c "import json; prompts=json.load(open('PROMPT-INDEX.json')); print([p for p in prompts if p['category']=='analysis'])"`。想验证提示词质量，运行 `./.claude/skills/prompt-refactor/scripts/validate-prompt.sh ./prompts/my-prompt.md`。所有提示词都遵循统一模板，含角色、上下文和交付标准。
📋 迷你测评
精选AI提示词集合，覆盖多种任务场景。亮点是“精选”属性，但简介未说明具体内容或数量。定位模糊，更像通用提示词包，适合随手查用。
📄 README 要点
这是一个生产就绪的AI提示词库，包含488个标准化提示词，全部通过11项质量校验，保证结构一致、交付物清晰、质量标准可衡量。能把ChatGPT、Claude等AI助手变成专家顾问。库里除了488+个覆盖47个类别的提示词，还有260+个带代码示例的Claude Code技能，以及7个安全、测试、格式化和CI/CD自动化钩子。每个提示词都遵循统一XML标签结构，方便解析。安装使用就是直接拿这些提示词模板去用，或者把技能装进Claude Code。适合需要大量高质量、结构化AI提示词做生产力工具的人，省去自己写提示词试错的时间。
查看英文原文
Useful AI Prompts Production Ready AI Prompt Library 488 production ready AI prompts, all following a standardized template with validated quality gates. Transform ChatGPT, Claude, and other AI assistants into expert consultants. Every prompt in this library passes 11 quality validation checks, ensuring consistent structure, clear deliverables, and measurable quality criteria. What's Inside Resource Type Count Description AI Prompts 488+ Standardized expert prompts across 47 categories Claude Code Skills 260+ Auto triggering capabilities with code examples Automation Hooks 7 Security, testing, formatting, and CI/CD automation Standardized Prompt Format Every prompt follows the same validated structure: [安装/使用命令见下方] XML Tag Structure The prompt section uses semantic XML tags for consistent parsing: Tag Purpose Content Expert identity Specific credentials, experience, reasoning approach Situation framing When used, success criteria, key assumptions Input specification Required vs optional inputs with defaults Process steps 3 7 numbered action steps Deliverable format Format, length, structure, must include elements Success measures Objective standards and anti patterns Hard boundaries Non negotiable limits on scope/format Quality Gates Every prompt passes these 11 validation checks: Gate Requirement metadata complete All required fields present (ID, Version, Category, Tags, Complexity, Interaction, Models) overview concise 3 sentences or fewer, no marketing fluff role specific Concrete expertise defined, not "I'll help you" inputs categorized Required vs optional distinguished with defaults task structured 3 7 clear numbered steps outputs specified Format + length + requirements for each deliverable criteria measurable Objective quality standards, not vague aspirations example realistic Input shows real usage (20 200 words) example concise Output demonstrates pattern (100 600 words) no duplication No repeated information across sections copy paste ready Prompt section
💬 评论（0）
交流使用体验、避坑建议；违规内容将被删除