Claude Code.
Anthropic's terminal-native coding agent — the reference for agent-driven engineering.
What it is.
Anthropic's terminal-native coding agent. Distinct from Claude on claude.com — Claude Code lives in the developer's shell, edits files in place, runs commands, manages git, and chains hooks to enforce policy. The reference deployment for agentic coding at Anthropic-aligned teams.
Where it fits.
Engineering organisations standardising on Claude as the substrate. Internal platform teams use Claude Code's hooks to enforce style, security, and review policy without third-party orchestration. Stripe, Adyen, and Block have public-record Claude Code deployments.
- Tightest integration with Anthropic models and capabilities
- Hooks framework enables policy-as-code without orchestrators
- Strong git and PR workflow built in
- Terminal-first UX excludes some collaborator roles
- Anthropic-only model lock — no GPT-5 or Gemini routing
Frequently asked.
Is Claude Code included with Claude Pro?
Yes. Both Pro and Max plans include Claude Code at quota tied to the plan.
Can Claude Code run unattended?
Yes. Background agent mode plus hooks allows long-running unattended workflows. Used in production for nightly refactors and dependency updates.
How does Claude Code compare to Cursor?
Cursor is a full IDE; Claude Code is terminal-first. Many teams run both — Cursor in the editor, Claude Code for unattended and hook-enforced workflows.
