MCP HubMCP Hub
返回技能列表

claude-cookbooks

2025Emma
更新于 Today
103 次查看
829
95
829
在 GitHub 上查看
aiapidesign

关于

This skill provides comprehensive code examples and tutorials for Claude API integration, including tool use, multimodal features, and RAG implementations. Use it when building Claude-powered applications, learning API usage, or implementing advanced patterns like AI agents. It offers practical guidance for developers working with Claude's capabilities across various integration scenarios.

快速安装

Claude Code

推荐
插件命令推荐
/plugin add https://github.com/2025Emma/vibe-coding-cn
Git 克隆备选方式
git clone https://github.com/2025Emma/vibe-coding-cn.git ~/.claude/skills/claude-cookbooks

在 Claude Code 中复制并粘贴此命令以安装该技能

技能文档

Claude Cookbooks Skill

Comprehensive code examples and guides for building with Claude AI, sourced from the official Anthropic cookbooks repository.

When to Use This Skill

This skill should be triggered when:

  • Learning how to use Claude API
  • Implementing Claude integrations
  • Building applications with Claude
  • Working with tool use and function calling
  • Implementing multimodal features (vision, image analysis)
  • Setting up RAG (Retrieval Augmented Generation)
  • Integrating Claude with third-party services
  • Building AI agents with Claude
  • Optimizing prompts for Claude
  • Implementing advanced patterns (caching, sub-agents, etc.)

Quick Reference

Basic API Usage

import anthropic

client = anthropic.Anthropic(api_key="your-api-key")

# Simple message
response = client.messages.create(
    model="claude-3-5-sonnet-20241022",
    max_tokens=1024,
    messages=[{
        "role": "user",
        "content": "Hello, Claude!"
    }]
)

Tool Use (Function Calling)

# Define a tool
tools = [{
    "name": "get_weather",
    "description": "Get current weather for a location",
    "input_schema": {
        "type": "object",
        "properties": {
            "location": {"type": "string", "description": "City name"}
        },
        "required": ["location"]
    }
}]

# Use the tool
response = client.messages.create(
    model="claude-3-5-sonnet-20241022",
    max_tokens=1024,
    tools=tools,
    messages=[{"role": "user", "content": "What's the weather in San Francisco?"}]
)

Vision (Image Analysis)

# Analyze an image
response = client.messages.create(
    model="claude-3-5-sonnet-20241022",
    max_tokens=1024,
    messages=[{
        "role": "user",
        "content": [
            {
                "type": "image",
                "source": {
                    "type": "base64",
                    "media_type": "image/jpeg",
                    "data": base64_image
                }
            },
            {"type": "text", "text": "Describe this image"}
        ]
    }]
)

Prompt Caching

# Use prompt caching for efficiency
response = client.messages.create(
    model="claude-3-5-sonnet-20241022",
    max_tokens=1024,
    system=[{
        "type": "text",
        "text": "Large system prompt here...",
        "cache_control": {"type": "ephemeral"}
    }],
    messages=[{"role": "user", "content": "Your question"}]
)

Key Capabilities Covered

1. Classification

  • Text classification techniques
  • Sentiment analysis
  • Content categorization
  • Multi-label classification

2. Retrieval Augmented Generation (RAG)

  • Vector database integration
  • Semantic search
  • Context retrieval
  • Knowledge base queries

3. Summarization

  • Document summarization
  • Meeting notes
  • Article condensing
  • Multi-document synthesis

4. Text-to-SQL

  • Natural language to SQL queries
  • Database schema understanding
  • Query optimization
  • Result interpretation

5. Tool Use & Function Calling

  • Tool definition and schema
  • Parameter validation
  • Multi-tool workflows
  • Error handling

6. Multimodal

  • Image analysis and OCR
  • Chart/graph interpretation
  • Visual question answering
  • Image generation integration

7. Advanced Patterns

  • Agent architectures
  • Sub-agent delegation
  • Prompt optimization
  • Cost optimization with caching

Repository Structure

The cookbooks are organized into these main categories:

  • capabilities/ - Core AI capabilities (classification, RAG, summarization, text-to-SQL)
  • tool_use/ - Function calling and tool integration examples
  • multimodal/ - Vision and image-related examples
  • patterns/ - Advanced patterns like agents and workflows
  • third_party/ - Integrations with external services (Pinecone, LlamaIndex, etc.)
  • claude_agent_sdk/ - Agent SDK examples and templates
  • misc/ - Additional utilities (PDF upload, JSON mode, evaluations, etc.)

Reference Files

This skill includes comprehensive documentation in references/:

  • main_readme.md - Main repository overview
  • capabilities.md - Core capabilities documentation
  • tool_use.md - Tool use and function calling guides
  • multimodal.md - Vision and multimodal capabilities
  • third_party.md - Third-party integrations
  • patterns.md - Advanced patterns and agents
  • index.md - Complete reference index

Common Use Cases

Building a Customer Service Agent

  1. Define tools for CRM access, ticket creation, knowledge base search
  2. Use tool use API to handle function calls
  3. Implement conversation memory
  4. Add fallback mechanisms

See: references/tool_use.md#customer-service

Implementing RAG

  1. Create embeddings of your documents
  2. Store in vector database (Pinecone, etc.)
  3. Retrieve relevant context on query
  4. Augment Claude's response with context

See: references/capabilities.md#rag

Processing Documents with Vision

  1. Convert document to images or PDF
  2. Use vision API to extract content
  3. Structure the extracted data
  4. Validate and post-process

See: references/multimodal.md#vision

Building Multi-Agent Systems

  1. Define specialized agents for different tasks
  2. Implement routing logic
  3. Use sub-agents for delegation
  4. Aggregate results

See: references/patterns.md#agents

Best Practices

API Usage

  • Use appropriate model for task (Sonnet for balance, Haiku for speed, Opus for complex tasks)
  • Implement retry logic with exponential backoff
  • Handle rate limits gracefully
  • Monitor token usage for cost optimization

Prompt Engineering

  • Be specific and clear in instructions
  • Provide examples when needed
  • Use system prompts for consistent behavior
  • Structure outputs with JSON mode when needed

Tool Use

  • Define clear, specific tool schemas
  • Validate inputs and outputs
  • Handle errors gracefully
  • Keep tool descriptions concise but informative

Multimodal

  • Use high-quality images (higher resolution = better results)
  • Be specific about what to extract/analyze
  • Respect size limits (5MB per image)
  • Use appropriate image formats (JPEG, PNG, GIF, WebP)

Performance Optimization

Prompt Caching

  • Cache large system prompts
  • Cache frequently used context
  • Monitor cache hit rates
  • Balance caching vs. fresh content

Cost Optimization

  • Use Haiku for simple tasks
  • Implement prompt caching for repeated context
  • Set appropriate max_tokens
  • Batch similar requests

Latency Optimization

  • Use streaming for long responses
  • Minimize message history
  • Optimize image sizes
  • Use appropriate timeout values

Resources

Official Documentation

Community

Learning Resources

Working with This Skill

For Beginners

Start with references/main_readme.md and explore basic examples in references/capabilities.md

For Specific Features

  • Tool use → references/tool_use.md
  • Vision → references/multimodal.md
  • RAG → references/capabilities.md#rag
  • Agents → references/patterns.md#agents

For Code Examples

Each reference file contains practical, copy-pasteable code examples

Examples Available

The cookbook includes 50+ practical examples including:

  • Customer service chatbot with tool use
  • RAG with Pinecone vector database
  • Document summarization
  • Image analysis and OCR
  • Chart/graph interpretation
  • Natural language to SQL
  • Content moderation filter
  • Automated evaluations
  • Multi-agent systems
  • Prompt caching optimization

Notes

  • All examples use official Anthropic Python SDK
  • Code is production-ready with error handling
  • Examples follow current API best practices
  • Regular updates from Anthropic team
  • Community contributions welcome

Skill Source

This skill was created from the official Anthropic Claude Cookbooks repository: https://github.com/anthropics/claude-cookbooks

Repository cloned and processed on: 2025-10-29

GitHub 仓库

2025Emma/vibe-coding-cn
路径: i18n/zh/skills/claude-cookbooks

相关推荐技能

content-collections

Content Collections 是一个 TypeScript 优先的构建工具,可将本地 Markdown/MDX 文件转换为类型安全的数据集合。它专为构建博客、文档站和内容密集型 Vite+React 应用而设计,提供基于 Zod 的自动模式验证。该工具涵盖从 Vite 插件配置、MDX 编译到生产环境部署的完整工作流。

查看技能

creating-opencode-plugins

该Skill为开发者创建OpenCode插件提供指导,涵盖命令、文件、LSP等25+种事件类型。它详细说明了插件结构、事件API规范及JavaScript/TypeScript实现模式,帮助开发者构建事件驱动的模块。适用于需要拦截操作、扩展功能或自定义AI助手行为的插件开发场景。

查看技能

sglang

SGLang是一个专为LLM设计的高性能推理框架,特别适用于需要结构化输出的场景。它通过RadixAttention前缀缓存技术,在处理JSON、正则表达式、工具调用等具有重复前缀的复杂工作流时,能实现极速生成。如果你正在构建智能体或多轮对话系统,并追求远超vLLM的推理性能,SGLang是理想选择。

查看技能

evaluating-llms-harness

测试

该Skill通过60+个学术基准测试(如MMLU、GSM8K等)评估大语言模型质量,适用于模型对比、学术研究及训练进度追踪。它支持HuggingFace、vLLM和API接口,被EleutherAI等行业领先机构广泛采用。开发者可通过简单命令行快速对模型进行多任务批量评估。

查看技能