Claude Code

Claude Code is Anthropic’s agentic coding tool for the terminal.

Overview

Type	CLI
Open Source	No
Best For	Agentic workflows, MCP

Key Features

• Agentic Coding — Reads, writes, executes autonomously
• Multi-file Editing — Understands project structure
• Terminal Commands — Runs build, test, lint
• MCP Support — External tool integration
• CLAUDE.md — Project context files

Installation

curl -fsSL https://claude.ai/install.sh | bash

Access Model

Claude Code uses Anthropic account access. Availability and plan details change frequently, so treat the vendor docs as the source of truth.

Workflow fit: best when you want a strong hosted terminal agent and are comfortable with Anthropic’s account and policy model.

Privacy

Setting	Value
Privacy Mode	Enterprise only
Training	ON by default (consumer)
Retention	30 days or 5 years
Jurisdiction	US

Privacy Warning

Consumer accounts (Free/Pro/Max) have training ON by default since August 2024. Enterprise accounts can disable this.

Getting Started

1. Install Claude Code
2. Authenticate with Claude account
3. Create CLAUDE.md in your project root
4. Run claude in your project directory

CLAUDE.md Example

# Project Context

## Commands
- `npm run dev` - Start dev server
- `npm test` - Run tests

## Gotchas
- Auth tokens in cookies, not localStorage

Tips

• Use /clear to reset context
• Create CLAUDE.md for project-specific context
• Use subagents for research: “Use subagents to investigate X”