anthropics/anthropic-cookbook

Features

• Prompt Engineering - Serves as a comprehensive guide for designing and refining prompts to optimize LLM performance.
• Agent Orchestration Frameworks - Provides a framework for orchestrating multi-step agentic workflows and coordinating multiple models.
• Agent Task Orchestrators - Implements patterns for coordinating multiple models to split complex tasks between high-reasoning and high-efficiency agents.
• Claude Integration Frameworks - Provides a framework for building applications using Claude's tool calling and agent coordination capabilities.
• Hierarchical Coordination - Provides a framework for coordinating multiple models in a hierarchy to split tasks between reasoning and efficiency agents.
• Retrieval-Augmented Generation - Implements patterns for grounding model responses by retrieving relevant data from vector databases.
• External Tool Integrations - Connects AI assistants to external functions, calculators, and databases to execute actions and retrieve real-time data.
• LLM Application Development - Demonstrates the creation of functional software using structured outputs, tool use, and agent patterns.
• Prompt Engineering Guides - Offers a collection of notebooks and recipes for implementing advanced prompting techniques and workflow patterns.
• Prompt Quality Scoring - Provides a system for using model-based feedback to quantify and score the clarity and robustness of prompts.
• RAG Context Retrieval - Implements the retrieval of relevant document segments from knowledge bases to ground LLM responses.
• Structured Output Enforcements - Enforces strict data models or schemas on LLM outputs to ensure reliable programmatic processing.
• Agentic Goal Decomposition - Implements patterns for using LLMs to recursively break high-level objectives into actionable sub-tasks for multi-agent coordination.
• Tool-Execution Loops - Orchestrates the request-response cycle between an AI model's tool request and the client's execution result.
• Retrieval Augmented Generation - Implements frameworks for enhancing model knowledge by integrating external data from vector databases and documents.
• Schema-Constrained Outputs - Forces model responses to conform to a provided JSON schema using prompt engineering for programmatic compatibility.
• RAG Implementation Guides - Provides step-by-step instructions and examples for building retrieval-augmented generation systems.
• LLM Tooling Integrations - Provides examples and guides for connecting LLMs to external functions, calculators, and databases.
• Prompt Optimizers - Offers methods for systematically refining and evaluating prompts to improve accuracy and reduce costs.
• Multimodal Data Processing - Provides technical patterns for extracting and interpreting information from images and PDF documents.
• Implementation Patterns - Ships technical demonstrations for processing visual information and parsing PDF documents using multimodal LLMs.
• PDF Document Analyzers - Combines text extraction and semantic search to analyze and process information within PDF documents.
• Prompt Caching - Implements prompt caching optimization techniques to improve performance and reduce operational costs.
• Prefix Caching - Demonstrates techniques for storing and sharing common prompt prefixes to reduce redundant computation and costs.
• Semantic Search - Demonstrates building search systems that understand query intent using embeddings and vector similarity.
• Semantic Vector Search - Uses mathematical distance between query and document embeddings to retrieve the most relevant text segments.
• Visual Content Analysis - Uses vision models to analyze images, charts, and forms to extract data and interpret visual content.
• Notebook-Based Experimentation - Uses interactive notebooks to combine executable code cells with documentation for demonstrating AI implementation patterns.
• LLM Evaluation - Provides a suite for systematically testing and scoring prompt quality using model-based evaluation.
• Agentic Engineering - Collection of code examples for building effective agentic workflows.

Star history