32dots HEIDELBERG AI
Curriculum › AI coding assistants › Claude Code course
EXTENDED COURSE

Claude Code

Claude Code is Anthropic's agentic coding tool: it runs in your terminal (also VS Code, JetBrains, Desktop, and the web), reads your whole project, makes coordinated multi-file edits, runs your code, and works with git — all from plain-English requests. This hands-on course takes you from your first terminal task to a pro workflow: plan mode, project memory in CLAUDE.md, slash commands, git, and cost control — then deeper into MCP, subagents, skills, hooks, and headless scripting.

1Lessons12step-by-step, ~90 min each

Get started

  1. 00 Start here: your first terminal task
  2. 01 Prompt-craft & iterating
  3. 02 Plan before you edit: Plan mode

Work like a pro

  1. 03 Project memory: CLAUDE.md
  2. 04 Slash commands & managing context
  3. 05 Git, commits & pull requests
  4. 06 Models, usage & cost

Go deep

  1. 07 Connect your tools: MCP
  2. 08 Delegate with subagents
  3. 09 Package workflows as Skills
  4. 10 Automate actions with Hooks
  5. 11 Scripting & CI: headless mode
2Cheat sheetcopy-ready expressions
3Examples3what people built

Internal tools & ops1

Claude Code Founder AI-native SME
Course starter

Add a REST endpoint + OpenAPI docs to an existing FastAPI app

In an existing FastAPI codebase, Claude Code adds a new `/reports/summary` GET endpoint: it reads the router pattern, creates the Pydantic schema, wires the database query, writes the route, adds tests, and updates the OpenAPI description — touching five files consistently.

The endpoint is live and tested within 20 minutes, and the generated OpenAPI spec is correct on the first try.

Try it yourself

Run `claude` in the project root: "Add a GET /reports/summary endpoint to the FastAPI app. Follow the pattern in app/routers/projects.py. Return aggregated row counts grouped by status. Add tests in tests/routers/."

Research & data tools1

Claude Code Scientist
Course starter

Backfill unit tests for a legacy analysis module

Point Claude Code at a 1 500-line Python analysis module with no tests. It reads the module, infers the expected behaviour of each function, and writes a pytest suite with edge-case coverage — all in the existing repo, no new scaffolding.

Coverage goes from 0 % to 87 % in one session, and the tests immediately surface two silent numerical bugs in the normalisation step.

Try it yourself

Run `claude` in the repo root, then: "Write pytest tests for src/analysis/plate_normalise.py. Match the existing pytest conventions in tests/. Run the suite and fix any failures."

Knowledge & docs1

Claude Code AI-native SME Founder
In the gallery

Migrate a Node.js utility library from JS to TypeScript

Claude Code renames files, adds type annotations, creates a tsconfig, updates the package.json build script, and fixes the resulting type errors across the entire library — iterating until `tsc --noEmit` passes with zero errors.

A 30-file library is fully typed in one session; downstream teams immediately benefit from autocomplete and compile-time safety.

Try it yourself

Run `claude` in the library root: "Convert this Node.js library to TypeScript. Rename .js → .ts, add strict type annotations, set up tsconfig.json, and iterate until tsc --noEmit passes."