Claude Code Documentation
Local reference for Claude Code CLI — every env var, settings key, slash command, tool, skill, hook, permission rule, and more — cross-referenced and organized into seventeen reference hubs.
Get Started
Section titled “Get Started” Getting Started Tutorials for new users — learning paths, first skill, first permission rule, first hook, feature gates guide.
.claude Directory Every file Claude Code reads from your project and home directory — CLAUDE.md, settings, skills, agents, memory, and more.
Use Claude Code
Section titled “Use Claude Code” CLI claude command-line flags, subcommands, invocation modes. Includes hidden flags from the source tree.
Commands Built-in slash commands — documented, internal source-tree finds, tombstones, experimental, and feature-flag-gated. 97 commands across 14 sections.
Keybindings Every shipped keyboard shortcut across 18 contexts — global, chat, selection, dialogs, diff — plus platform quirks and custom-override config.
Tools Every tool available to Claude — filesystem, shell, network, MCP, orchestration. 44 tools across 16 sections.
Configure & Secure
Section titled “Configure & Secure” Settings settings.json key reference — auth, model, permissions, sandbox, plugins, MCP, hooks, UI, worktree. 119 keys across 19 sections.
Environment Variables Every environment variable recognized by Claude Code — 500+ vars across 21 sections.
Permissions Rule grammar, evaluation model, permission modes, scope precedence, and auto-mode classifiers.
Hooks Event-hook catalog — 26 events × 4 handler types × matcher syntax, plus the SSRF security model.
Build & Extend
Section titled “Build & Extend” Skills Built-in and custom skills — frontmatter schema (29 fields), bundled-skills catalog, and how to add your own.
Plugins plugin.json manifest reference, directory layout, lifecycle, marketplace concept, and bundled-plugin catalog.
Agent SDK TypeScript/Python SDK for building custom agents — session lifecycle, message protocol, streaming events, tool bridging.
Agents Subagent overview — how Claude spawns and coordinates subagents, definition methods, and orchestration patterns.
Reference
Section titled “Reference” Sessions Transcript storage, 30-day cleanup, auto-compaction, context caching, 17 entry types, and session resume.
Memory Auto-memory layout, MEMORY.md caps, 200-file scan, type classification, and team + agent memory scopes.
Changelog Release history, silent model migrations, and deprecation notices.