2.7 — Reading Documentation

Documentation is how engineers communicate how their software works. Learning to read docs is a fundamental builder skill — and a critical AI safety habit.

Five Types of Documentation

Type	What It Contains	When You Need It
README	Project overview, setup instructions, basic usage	First time encountering any project
API Reference	Every function, parameter, and return value	When you need to know exactly how to call something
Tutorial	Step-by-step guide for a specific task	When learning something new from scratch
Changelog	What changed between versions	When updating dependencies or diagnosing breaking changes
Inline comments	Notes inside the code itself	When reading source code to understand what a function does

The Critical Skill

Before asking AI to explain a library or tool, check if it has documentation. AI can hallucinate API details. Documentation can’t.

This is not a knock on AI — it’s a practical workflow. AI is trained on older data and may not know the current API for a library you’re using. When AI says “call it like this” and it doesn’t work, the documentation is the ground truth.

The workflow:

1. Open the library’s official documentation
2. Find the specific function or feature you need
3. Read the parameters and example code
4. Ask AI to help you apply it to your specific situation — with the documentation in context

Giving AI the relevant documentation in your prompt produces dramatically more accurate answers than asking from memory.

Where to Find Documentation

• README — usually at the root of any GitHub repository
• API Reference — typically at docs.[library].com or in the repository’s /docs folder
• Changelogs — CHANGELOG.md in the repository, or the “Releases” page on GitHub
• Inline comments — open the source file in your IDE and read

The Anthropic documentation (for Claude and Claude Code) and MDN Web Docs (for web technologies) are two of the most reliable references you’ll use repeatedly.

---

Next: 2.8 — Vocabulary | Phase overview: Phase 2