Building with the Claude API
by Anthropic
Anthropic's own 84-lesson path from a first API call to production evals, RAG, MCP servers and agent workflows.
Overview
Building with the Claude API is the largest course in Anthropic Academy's catalogue — 84 lessons, more than eight hours of video and ten quizzes, free after creating an account, and repeatedly ranked first among the Academy's roughly fourteen courses by outside reviewers. It runs in ten sections. It opens on API access: getting a key, making a request, multi-turn conversations, system prompts, temperature, response streaming and structured data, each with its own coding exercise. Section three is prompt evaluation, placed unusually early — a typical eval workflow, generating test datasets, running the eval, and both model-based and code-based grading — so measurement precedes technique. Prompt engineering follows, covering clarity, specificity, XML tag structure and examples. A long tool-use section builds a project through tool functions, tool schemas, handling message blocks, sending tool results, multi-turn and multi-tool loops, fine-grained tool calling, and the built-in text-edit and web-search tools. The RAG section builds the full flow: chunking strategies, embeddings, BM25 lexical search and a multi-index pipeline. A features section covers extended thinking, image and PDF support, citations, prompt caching and the code execution and Files API. Section eight builds an MCP server and client from scratch — tools, resources, prompts, the inspector — and section ten separates workflows from agents across parallelization, chaining and routing. A final assessment issues the certificate.
At a Glance
- Topic
- Models
- Level
- Intermediate
- Format
- Course
- Cost
- Free
- Duration
- ~8+ hours of video across 84 lessons and 10 quizzes, self-paced
- Provider
- Anthropic
- Hands-on
- Yes — code/exercises
- Certificate
- Available
What You’ll Learn
- ✓Call the Claude API with streaming, system prompts and structured data outputs
- ✓Run a full evaluation workflow using model-based and code-based grading
- ✓Implement tool use end to end, including multi-turn loops and multiple tools
- ✓Build a RAG pipeline with chunking, embeddings, BM25 and multi-index retrieval
- ✓Use prompt caching, extended thinking, citations and the Files API deliberately
- ✓Write an MCP server and client exposing tools, resources and prompts
- ✓Choose between chaining, routing, parallelization workflows and a true agent loop
Highlights
- •The largest course on Anthropic Academy — 84 lessons and ten quizzes — and ranked the single most valuable resource on the platform by independent reviewers
- •Teaches prompt evaluation in section three, before prompt engineering, which is the reverse of almost every other API course
- •Covers current platform surface area: fine-grained tool calling, extended thinking, prompt caching rules, code execution and the Files API, and the built-in web search tool
- •Builds an MCP server and client from scratch, including the inspector, rather than only explaining MCP concepts
- •Free with an email sign-up, and the completion certificate carries a permanent verification URL you can attach to LinkedIn
Who It’s For
Best For
- ✓Backend or full-stack engineers integrating Claude into a product
- ✓Developers who want Anthropic's own account of tool use and MCP
- ✓Teams standing up their first LLM evaluation workflow
- ✓Engineers moving from prompt experiments to a production RAG or agent system
Prerequisites
- •Proficiency writing Python
- •Basic knowledge of handling JSON data
- •No prior Claude experience — the course starts at getting an API key
FAQ
What is Building with the Claude API?
Anthropic Academy's flagship developer course: 84 free lessons and more than eight hours of video taking a Python developer from a first Claude API request through to production RAG, a hand-built MCP server and agent workflows. It is for backend and full-stack engineers integrating Claude into a product who want the vendor's own account of tool use, prompt caching, extended thinking and evaluation, with ten quizzes and a verifiable certificate.
Is Building with the Claude API free?
Building with the Claude API is free to access.
What level is Building with the Claude API for?
Building with the Claude API is aimed at a intermediate audience. Recommended background: Proficiency writing Python, Basic knowledge of handling JSON data, No prior Claude experience — the course starts at getting an API key.
How long does Building with the Claude API take?
Expect roughly ~8+ hours of video across 84 lessons and 10 quizzes, self-paced. Most learners work through it at their own pace.
What will I learn from Building with the Claude API?
You'll learn: Call the Claude API with streaming, system prompts and structured data outputs; Run a full evaluation workflow using model-based and code-based grading; Implement tool use end to end, including multi-turn loops and multiple tools; Build a RAG pipeline with chunking, embeddings, BM25 and multi-index retrieval; Use prompt caching, extended thinking, citations and the Files API deliberately; Write an MCP server and client exposing tools, resources and prompts; Choose between chaining, routing, parallelization workflows and a true agent loop.
Topics
Sources
This page was written from 3 sources, 2 on domains other than anthropic.skilljar.com.