cat learning-paths.md
Free structured courses from Anthropic. Complete them in order for the best learning experience — or jump to the topic you need.
REQUIREMENTS:
- PYTHON — Basic Python knowledge
- API_KEY — An Anthropic API key (get one here)
- JUPYTER — Jupyter Notebook or Google Colab
Anthropic API Fundamentals
The essentials of working with the Claude SDK. Learn how to get an API key, understand the messages format, explore the Claude model family, configure model parameters, stream responses, and work with vision/multimodal prompting.
TOPICS COVERED:
- >Getting started & API key setup
- >Messages format
- >Claude model family
- >Model parameters (temperature, max_tokens, etc.)
- >Streaming responses
- >Vision & multimodal prompting
Prompt Engineering Interactive Tutorial
A comprehensive 9-chapter guide to prompt engineering, from basic structure through advanced techniques. Covers role prompting, separating data from instructions, formatting output, chain-of-thought reasoning, few-shot examples, and avoiding hallucinations.
TOPICS COVERED:
- >Basic prompt structure & being clear
- >Assigning roles to Claude
- >Separating data from instructions
- >Formatting output (JSON, XML, markdown)
- >Chain-of-thought reasoning
- >Few-shot examples
- >Avoiding hallucinations
- >Complex prompts from scratch
Real World Prompting
Apply prompting techniques to complex, real-world domains. Work through medical, customer support, and call summarization use cases using the iterative prompt engineering process.
TOPICS COVERED:
- >Prompting technique recap
- >Medical domain prompts
- >Iterative prompt engineering process
- >Call summarization
- >Customer support AI
Prompt Evaluations
Learn how to measure and evaluate prompt quality in production. Covers human-graded evals, code-graded evals, classification evaluations, and integration with the promptfoo evaluation framework.
TOPICS COVERED:
- >Introduction to evaluations
- >Human-graded evals (Workbench)
- >Code-graded evaluations
- >Classification evaluations
- >promptfoo framework integration
- >Custom grader functions
- >Model-as-judge evaluations
Tool Use (Function Calling)
Everything you need to implement tool use successfully. Learn how Claude calls external functions, how to structure tool definitions, control tool choice, and build multi-tool chatbots.
TOPICS COVERED:
- >Tool use overview & architecture
- >Building your first tool
- >Structured outputs via tool use
- >Complete request/response workflow
- >Controlling tool choice
- >Multi-tool chatbot
ALL COURSES ARE FREE
These courses are maintained by Anthropic and hosted on GitHub. They use Claude 3 Haiku by default to keep API costs low, but you can use any Claude model. Run the notebooks locally or on Google Colab.
VIEW_ALL_ON_GITHUB