Reading Anthropic's docs efficiently

The Anthropic docs are deep but non-linear. Here are three reading orders we recommend depending on your background.

Order A — Total beginner

1. Claude 101 (Skilljar)
2. Building with Claude API (Skilljar)
3. Prompt engineering overview
4. Agent SDK overview
5. Claude Code intro

Order B — Backend engineer new to LLMs

1. Agent SDK overview
2. Tool use guide
3. Prompt caching deep-dive
4. MCP spec
5. Engineering blog: "Building Agents with Claude Agent SDK"

Order C — Frontend / full-stack already shipping with Claude

1. MCP spec
2. Claude Code hooks + skills docs
3. Agent evaluation patterns
4. Context management deep-dive
5. Production rollout playbook

Whatever order, read with a notebook open. Tab-skimming feels productive and rarely is.