anthropic-claude-skills
- 评测生成时间(北京时间)
- 本报告引擎
- v3.10.0
- 当前引擎
- v3.16.0
本报告与当前引擎使用不同规则;原分数不会自动更新,不同版本的分数不宜直接对比。
进入后确认来源与额度,提交才会创建任务。
综合采用结论
存在需要人工复核的风险或证据不足
- 基础评测完成+25/25确定性评分与静态安全扫描已完成
- README 有效证据+7/252,757 个去重后的有效字符
- 独立证据来源+4/201 类非重复证据,重复文件不叠加
- 仓库元数据+10/10已取得仓库状态与采用数据
- 活跃记录+5/5已取得最近提交时间
- AI 复核+15/15已完成结构化 AI 证据复核
公开证据不足,暂不绘图
当前材料不足以同时核实至少 2 个步骤或组件及 1 条关系。为避免臆测,报告保留文字证据,不生成流程图、时序图或架构图。
- README中列出四个技能:p5js-interactive-graphics、phaser-game-dev、prd-builder、transcript-to-interactive
- 每个技能有'Best for'和'Why it helps'描述
- 使用方式:'Copy them directly into Claude'
- 声明'Not production-ready frameworks'
- 有效 README
- 输入、参数或工具说明
- 输出或结果说明
- 结构化章节
- 未发现已知高风险模式
- 缺少问题与用途描述
- 缺少安装或接入步骤
- 缺少可执行示例
- 缺少每个技能的具体提示词或示例
- 无安装或接入步骤
希望快速获取技能灵感的开发者、需要p5.js或Phaser开发辅助的用户、早期产品PRD构建、将转录内容转为互动体验
需要详细配置和示例的生产环境、需要明确错误处理或排障的场景
也有自己的公开项目?先看完证据,再用当前规则生成独立报告。
评测我的项目 →静态扫描不是安全保证,生产接入前仍应人工复核权限和数据边界。
- 01补充问题与用途描述
- 02补充安装或接入步骤
- 03补充可执行示例
方法、证据与局限展开收起
GitHub Repository API
1 个文件 · 2,984 字符
v3.10.0 · AI 复核已启用(deepseek-chat)
- 静态评测不会安装或执行项目代码
- 安全扫描基于高信号文件与已知模式,不能替代人工审计
- 流行度只反映采用程度,不代表安全或工程质量
30 天热度趋势
README
Claude Skills Toolkit 🧠⚡
A small set of opinionated, reusable Claude skills built while shipping real products, games, and experiments.
Instead of prompting Claude from scratch every time, I started treating it like a junior teammate with clearly defined responsibilities.
These skills are the result of that approach.
This is not a prompt dump.
It’s a builder’s toolkit optimized for speed, clarity, and iteration.
Why this exists
Most people use AI like this:
“Let me explain everything again and hope it works.”
I wanted:
- Faster iteration
- Less cognitive load
- Consistent output quality
- Reusable thinking, not just reusable text
So I started formalizing the prompts and structures I kept reusing.
This repo is that toolbox.
Included Skills
🎨 p5js-interactive-graphics
A Claude skill for generating and iterating on creative, interactive p5.js sketches.
Best for:
- Interactive visuals
- Creative coding experiments
- Generative art with user interaction
Why it helps:
- Avoids boilerplate explanations
- Encourages visual-first thinking
- Produces sketch-ready code quickly
🎮 phaser-game-dev
A Claude skill focused on Phaser-based game development, iteration, and refactoring.
Best for:
- Rapid game prototypes
- Power-ups, mechanics, and UI logic
- Config-driven game design
Why it helps:
- Knows common Phaser patterns
- Optimized for iteration, not tutorials
- Thinks in systems (state, config, scaling)
📄 prd-builder
An opinionated PRD builder for shipping-focused product teams.
Best for:
- Early-stage products
- Hackathons and MVPs
- Founder-led development
What it intentionally avoids:
- Enterprise fluff
- Over-documentation
- Vague vision-only PRDs
Philosophy:
Optimize for speed to prototype, not paperwork.
🎥 transcript-to-interactive
Turns long-form transcripts into interactive experiences instead of passive text.
Best for:
- YouTube videos
- Podcasts
- Courses and explainers
Examples:
- Transcript → quiz
- Transcript → reveal game
- Transcript → interactive learning flow
This skill bridges content → product → interaction.
How to use this repo
Each skill folder contains:
- A clear description of the skill
- Example prompts
- Suggested use cases
- Notes on when not to use it
You can:
- Copy them directly into Claude
- Modify them for your workflow
- Use them as inspiration to build your own skills
What this is not
- ❌ A replacement for thinking
- ❌ A no-code magic button
- ❌ Production-ready frameworks
These are evolving tools, not finished products.
Who this is for
- Developers using AI daily
- Indie hackers & founders
- Designers who code
- Anyone shipping fast with taste
Final note
If you’re treating AI as a collaborator instead of a chatbot, you’ll probably find this useful.
Feel free to fork, adapt, and break things.