getting-started-doc

This skill provides the complete official documentation for getting started with Claude Code — installation, authentication, quickstart, how the tool works, supported platforms and integrations, team rollout resources, and a full glossary.

Quick Reference

Installation Methods

| Method | Command / Action | Auto-updates | | :--- | :--- | :--- | | Native (macOS/Linux/WSL) | curl -fsSL https://claude.ai/install.sh \| bash | Yes | | Native (Windows PowerShell) | irm https://claude.ai/install.ps1 \| iex | Yes | | Homebrew (stable) | brew install --cask claude-code | No (manual: brew upgrade claude-code) | | Homebrew (latest) | brew install --cask claude-code@latest | No (manual: brew upgrade claude-code@latest) | | WinGet | winget install Anthropic.ClaudeCode | No (manual: winget upgrade Anthropic.ClaudeCode) | | npm | npm install -g @anthropic-ai/claude-code | No (manual: npm install -g @anthropic-ai/claude-code@latest) | | apt (Debian/Ubuntu) | See setup reference for signed-repo steps | No | | dnf (Fedora/RHEL) | See setup reference for signed-repo steps | No | | apk (Alpine Linux) | See setup reference for signed-repo steps | No |

System requirements: macOS 13+, Windows 10 1809+ / Server 2019+, Ubuntu 20.04+, Debian 10+, Alpine 3.19+; 4 GB RAM, x64 or ARM64.

Verify install: claude --version or claude doctor.

Essential CLI Commands

Shell commands (run from terminal)

| Command | What it does | | :--- | :--- | | claude | Start interactive mode | | claude "task" | Run a one-time task | | claude -p "query" | Non-interactive: run query, print result, exit | | claude -c | Continue most recent conversation in current directory | | claude -r | Resume a previous conversation (shows picker) | | claude --model <name> | Start with a specific model | | claude update | Apply update immediately | | claude setup-token | Generate a long-lived OAuth token for CI | | claude doctor | Diagnose installation and config issues |

Session commands (typed inside a running session)

| Command | What it does | | :--- | :--- | | /help | Show available commands | | /clear | Clear conversation history, start new session | | /resume | Resume or fork a previous session | | /login / /logout | Switch accounts or re-authenticate | | /init | Generate CLAUDE.md from your project structure | | /model | Switch models mid-session | | /context | Show what's using context window space | | /compact | Manually trigger context compaction | | /plan | Enter plan mode (review before edit) | | /goal <condition> | Keep Claude working until condition is met | | /goal | Check current goal status | | /goal clear | Clear an active goal | | /rewind | Restore to an earlier checkpoint | | /exit or Ctrl+D | Exit Claude Code |

Authentication Methods (Precedence Order)

| Priority | Method | When to use | | :---: | :--- | :--- | | 1 | Cloud provider env vars (CLAUDE_CODE_USE_BEDROCK, etc.) | Amazon Bedrock, Google Vertex AI, Microsoft Foundry | | 2 | ANTHROPIC_AUTH_TOKEN | LLM gateway / proxy with bearer tokens | | 3 | ANTHROPIC_API_KEY | Direct Anthropic API access | | 4 | apiKeyHelper script | Dynamic/rotating credentials from a vault | | 5 | CLAUDE_CODE_OAUTH_TOKEN | CI pipelines (generated via claude setup-token) | | 6 | Subscription OAuth (from /login) | Default for Pro, Max, Team, Enterprise |

Credential storage: macOS Keychain; ~/.claude/.credentials.json (Linux, mode 0600); %USERPROFILE%\.claude\.credentials.json (Windows).

Supported Account Types

| Account | Notes | | :--- | :--- | | Claude Pro / Max | Individual subscription at claude.com/pricing | | Claude for Teams | Self-service, centralized billing, admin tools | | Claude for Enterprise | Adds SSO, domain capture, role-based permissions, managed policy | | Claude Console | API-based billing; assign "Claude Code" or "Developer" roles | | Amazon Bedrock | Set CLAUDE_CODE_USE_BEDROCK=1 + AWS credentials | | Google Vertex AI | Set CLAUDE_CODE_USE_VERTEX=1 + GCP credentials | | Microsoft Foundry | Set CLAUDE_CODE_USE_FOUNDRY=1 + Azure credentials |

The Agentic Loop

Claude works through three phases per task: gather context → take action → verify results, repeating until done. Each tool use feeds into the next step.

Built-in tool categories

| Category | What Claude can do | | :--- | :--- | | File operations | Read, edit, create, rename files | | Search | Find files by pattern, search content with regex | | Execution | Run shell commands, tests, git operations | | Web | Search the web, fetch documentation | | Code intelligence | Type errors, definitions, references (requires plugins) |

Permission Modes (cycle with Shift+Tab)

| Mode | Behavior | | :--- | :--- | | default | Asks before file edits and shell commands | | acceptEdits | File edits and common filesystem commands flow without asking; other commands still prompt | | plan | Explores and proposes a plan; no source file edits until approved | | auto | Background safety classifier approves most actions (research preview) |

Platforms

| Platform | Best for | Key extras | | :--- | :--- | :--- | | CLI (Terminal) | Terminal workflows, scripting, remote servers | Full feature set, Agent SDK, computer use (macOS Pro/Max) | | Desktop app | Visual review, parallel sessions | Diff viewer, Dispatch, computer use | | VS Code | Working inside VS Code | Inline diffs, integrated terminal, file context | | JetBrains | IntelliJ, PyCharm, WebStorm, etc. | Diff viewer, selection sharing | | Web (claude.ai/code) | Long-running tasks, offline continuation | Anthropic-managed cloud, continues after disconnect | | Mobile (iOS/Android) | Starting/monitoring tasks away from desk | Cloud sessions, Remote Control, Dispatch |

Integrations: Chrome (browser automation), GitHub Actions, GitLab CI/CD, Code Review (auto PR review), Slack (@Claude mentions).

Remote access options

| Option | How it works | | :--- | :--- | | Remote Control | Drive a local session from browser or phone | | Dispatch | Message a task from mobile, Desktop session spawns | | Channels | Push events from Telegram/Discord/iMessage into a session | | Scheduled tasks | /loop (in-session), Desktop scheduled tasks, or cloud Routines |

Update & Release Channels

| Channel | Description | Set via | | :--- | :--- | :--- | | latest (default) | Every release as it ships | "autoUpdatesChannel": "latest" in settings.json | | stable | ~1 week behind, skips regressions | "autoUpdatesChannel": "stable" in settings.json |

Use minimumVersion in settings.json to pin a version floor. Disable auto-updates with "DISABLE_AUTOUPDATER": "1" in env.

/goal Command

/goal requires Claude Code v2.1.139+. It wraps a session-scoped Stop hook: after each turn, a small fast model checks the condition; if not met, Claude continues automatically.

| Usage | Effect | | :--- | :--- | | /goal <condition> | Set a goal; starts working immediately | | /goal | Show status (condition, turns, tokens, evaluator reason) | | /goal clear | Remove active goal before condition is met |

Writing effective conditions: state one measurable end state, a check Claude can run (e.g., "npm test exits 0"), and any constraints. Conditions up to 4,000 characters. Goals survive --resume but reset turn count.

Key Glossary Terms

| Term | Definition | | :--- | :--- | | Agentic loop | Gather context → take action → verify results, repeated until done | | Agentic harness | The tools, context management, and execution layer around the Claude model | | CLAUDE.md | Markdown file of persistent project instructions loaded every session | | Auto memory | Notes Claude writes for itself, stored per-repo under ~/.claude/projects/ | | Compaction | Auto-summarization when context window fills; CLAUDE.md and auto memory survive | | Checkpoint | Per-prompt restore point; press Esc twice or /rewind to revert | | Skill | SKILL.md file adding instructions or a workflow Claude can invoke | | Hook | Handler (command/HTTP/MCP/prompt/agent) that fires at lifecycle points | | Subagent | Specialized agent in its own context window with delegated task | | Plugin | Bundle of skills, hooks, subagents, and MCP servers as an installable unit | | Session | A conversation tied to a directory with its own context window | | Turn | One complete Claude response, including all tool calls within it | | Plan mode | Permission mode: Claude proposes changes before editing any source file | | Verification loop | A check Claude runs after each attempt to confirm work is actually done | | Non-interactive mode | -p flag: runs a single prompt and exits (formerly "headless mode") | | Bare mode | --bare: skips hooks, skills, plugins, MCP, auto memory, CLAUDE.md | | Teleport | /teleport pulls a cloud session into your local terminal | | Surface | Any access point: CLI, VS Code, JetBrains, Desktop, or claude.ai | | Managed settings | Org-wide settings enforced at OS-level paths; users cannot override | | Settings layers | Precedence: managed policy > CLI args > .local.json > project > user |

Team Rollout Quick Reference

Pre-launch checklist (from Communications Kit)

• #claude-code channel created and linked in announcement
• Install command tested in your environment (proxy/firewall check)
• Security/data-handling link ready (data-usage, security)
• One concrete first task from your own codebase chosen
• Named owner for the channel for the first 48 hours
• Executive sponsor lined up to send or co-sign

Champion role behaviors: share discoveries publicly, answer questions with actual prompts used, establish lightweight recurring habits (weekly show-and-tell, #claude-code channel).

High-leverage starter prompts

| Task | Prompt template | | :--- | :--- | | Fix a bug | "the tests in [file] are failing, figure out why and fix it" | | Understand code | "walk me through how [module] works, then tell me where the entry point is" | | Safe refactor | "refactor [module] to [goal], use plan mode so I can review first" | | Write tests | "write tests for [file] that cover the edge cases around [scenario]" | | Review before commit | "look at my working diff and tell me what looks risky" | | Open a PR | "fix [issue], write a conventional commit, and open a PR with a summary" |

Full Documentation

For the complete official documentation, see the reference files:

• Overview — What Claude Code is, all interfaces, and what you can do with it
• Quickstart — Step-by-step walkthrough of your first session
• Advanced Setup — System requirements, installation methods, updates, uninstallation, binary integrity
• Authentication — Account types, team setup, credential management, long-lived tokens
• How Claude Code Works — Agentic loop, built-in tools, sessions, checkpoints, permissions, working effectively
• Platforms and Integrations — Platform comparison, integrations, remote access options
• Keep Claude Working Toward a Goal — The /goal command, evaluation mechanics, non-interactive use
• Champion Kit — Playbook for engineers driving internal adoption
• Communications Kit — Launch announcements, drip campaigns, FAQ responses for org rollouts
• Glossary — Definitions for all Claude Code terminology

Sources

• Overview: https://code.claude.com/docs/en/overview.md
• Quickstart: https://code.claude.com/docs/en/quickstart.md
• Advanced Setup: https://code.claude.com/docs/en/setup.md
• Authentication: https://code.claude.com/docs/en/authentication.md
• How Claude Code Works: https://code.claude.com/docs/en/how-claude-code-works.md
• Platforms and Integrations: https://code.claude.com/docs/en/platforms.md
• Keep Claude Working Toward a Goal: https://code.claude.com/docs/en/goal.md
• Champion Kit: https://code.claude.com/docs/en/champion-kit.md
• Communications Kit: https://code.claude.com/docs/en/communications-kit.md
• Glossary: https://code.claude.com/docs/en/glossary.md