A collection of code examples and guides for building with the Claude API, maintained as Jupyter notebooks with copy-able Python snippets developers can adapt into their own projects. It targets developers who already have a Claude API key and want worked examples rather than reference docs; the concepts carry over to any language that can call the API.
Recipes are grouped by theme: core capabilities (classification, retrieval-augmented generation, summarization), tool use and integration (a customer-service agent, calculator integration, SQL queries), third-party integrations (Pinecone vector search, Wikipedia, web pages, Voyage AI embeddings), and multimodal work with vision (chart and form reading, image-generation via Stable Diffusion). An advanced section covers using Haiku as a sub-agent under Opus, PDF upload and summarization, automated prompt evaluations, JSON mode, moderation filters, and prompt caching. The project is community-driven and accepts contributed guides.
