Guide
Claude Code docs: where to start
Harrowtech.ai · last checked 13 October 2026
Anthropic's index of the Claude Code documentation listed 222 English pages on 13 October 2026. This guide says which page answers which question, and in what order a newcomer should read them.
Harrowtech.ai is independent of Anthropic. Every outside fact below was read on Anthropic's own pages on 13 October 2026. Where a page said nothing, this guide says nothing.
Where the docs live
The official Claude Code docs are at code.claude.com/docs. Anthropic also publishes a plain-text index of every page at code.claude.com/docs/llms.txt. That index groups the pages under Getting started, Build with Claude Code, Plugins, Administration, Configuration, Reference, Agent SDK, What’s New and Resources. It also links indexes in eleven other languages (French, German, Italian, Japanese, Spanish, Korean, Simplified Chinese, Traditional Chinese, Russian, Indonesian and Brazilian Portuguese), each listed at 222 pages.
The descriptions below are Anthropic’s own one-line summaries from that index, shortened. Page names can change, so if a link moves, use the index.
A reading order for a newcomer
Read these in this order. The first five are enough to be productive.
| Overview | What Claude Code is and where it runs: terminal, IDE, desktop, browser. Start here if you have not decided yet. Our plain-English version is What is Claude Code? |
|---|---|
| Quickstart | Install, sign in and make your first code change. It also lists the shell commands claude, claude -p, claude -c and claude -r. |
| How Claude Code works | The agentic loop, the built-in tools, sessions, the context window, checkpoints and permissions. |
| Memory | Persistent instructions through CLAUDE.md, and auto memory. See our guide to CLAUDE.md and projects. |
| Best practices | Tips for environment setup and parallel sessions. Our guide to it is Claude Code best practices. |
| Common workflows | Step-by-step guides for everyday tasks. |
| Prompt library | Copy-and-paste prompts by task and role. |
Find the page by task
Setting up
| Installing it | Setup: system requirements, platform installs, versions and uninstall. See also our install guide. |
|---|---|
| A failed install | Troubleshoot installation: installation, PATH, permission and login errors. |
| Signing in | Authentication: login for individuals, teams and organisations. Ours: API key and login. |
| Never used a terminal | The quickstart points to a terminal guide. |
| Working without the terminal | Desktop quickstart, VS Code, JetBrains and web quickstart. The platforms page compares them all. |
Using it day to day
| Slash commands and shortcuts | Commands: built-in commands and bundled skills. Keyboard shortcuts are on Interactive mode. Ours: commands and shortcuts. |
|---|---|
Flags for claude | CLI reference: commands and flags. |
| Resuming yesterday’s work | Sessions: naming, resuming, branching and exporting conversations. |
| Undoing a change | Checkpointing: tracking, rewinding and summarising edits and conversation. |
| Slow, or forgetting things | Context window, an interactive simulation of how the context fills, and Costs for token tracking and spend limits. |
Controlling and extending it
| What it asks permission for | Permission modes and Permissions, for fine-grained allow and deny rules. Security covers safeguards and safe use. |
|---|---|
| Where settings go | Settings for files, scopes and precedence, and Settings reference for every settings.json key. |
| Choosing an extension | Extend Claude Code explains when to use CLAUDE.md, skills, subagents, hooks, MCP or plugins. |
| Skills, hooks, MCP, subagents | Skills, Hooks guide, MCP quickstart and Subagents. Ours: Claude Code agents. |
Files in .claude | The .claude directory: files Claude Code reads from .claude/ and ~/.claude. |
When something is wrong
| It runs badly | Troubleshooting: high CPU or memory, hangs, auto-compact thrashing and search issues. |
|---|---|
| CLAUDE.md or settings ignored | Debug your config. |
| An error message | Errors: runtime error messages and fixes. |
| An unfamiliar term | Glossary: agentic loop, compaction, hooks, subagents, MCP and more. |
| Recent changes | Changelog for release notes by version, and What’s new for a weekly digest. |
Pages for teams and builders
- Rolling it out at work. Admin setup is the decision map for deploying Claude Code in an organisation. Managed settings, Data usage and the Enterprise deployment overview sit under Administration, with a page each for Amazon Bedrock, Google Cloud’s Agent Platform and Microsoft Foundry.
- Automating it. Headless covers programmatic use from the CLI, Python or TypeScript. GitHub Actions and GitLab CI/CD cover pipelines.
- Building your own agents. The Agent SDK overview describes Claude Code as a library for production agents.
- Large repositories. Large codebases covers monorepo setup with nested CLAUDE.md files.
A tip from the docs themselves
Anthropic’s “How Claude Code works” page says Claude Code can teach you how to use it. You can ask it questions such as “how do I set up hooks?” or “what’s the best way to structure my CLAUDE.md?”. The quickstart makes the same point and adds that /help is available inside a session. The quickstart also links Claude Code 101 and other free self-paced courses on Claude Academy, and a Discord server.
What this guide does not cover
It maps the pages. It does not summarise them all, and it leaves out the Plugins and Agent SDK sections beyond the overview, the Mods pages, and the gateway and self-hosted environment pages. The docs change often, so the index is the place to confirm what exists today. For a plain-English overview first, see What is Claude Code?. Every guide on this site is listed on the Guides page.
Sources
- Claude Code documentation index, read 13 October 2026
- Overview, Quickstart, How Claude Code works and Platforms, all read 13 October 2026