Claude Code

Claude Code is Anthropic's coding agent. It runs where developers already work — the terminal, the IDE, CI, and the web — and operates on real repositories: reading code, editing files across a project, running tests and builds, and committing the result.

What sets it apart

• Whole-codebase understanding. It explores a repository on demand rather than requiring you to paste context, so it can answer questions about unfamiliar code as easily as it can change it.
• Agentic execution. It runs commands, reads failures, and iterates until tests pass, instead of handing back a snippet for you to try.
• Surfaces to match the task. A terminal CLI for interactive work, IDE extensions for VS Code and JetBrains, a web and mobile experience for delegating tasks, and a GitHub integration that reviews PRs and responds to comments.
• Extensible. Skills, subagents, hooks, MCP servers, and slash commands let teams encode their own conventions so the agent follows house style.
• Enterprise deployment. Available directly, or through Amazon Bedrock, Google Vertex AI, and Microsoft Foundry.

Who uses it

Individual developers on Claude Pro and Max plans, and engineering organisations on team and enterprise plans, typically for refactors, test coverage, migrations, incident debugging, and code review — work that is well specified but tedious to do by hand.