What it does
The page presents Claude Code as a way to hand Claude a bug fix, a test or a long migration, then steer and review the work from your terminal, IDE, Slack or the web. It is available for macOS, Linux and Windows, with further ways to use it on iOS, Android, GitHub, VS Code, JetBrains and Slack. According to the FAQ, it plans, writes code, runs tests and opens pull requests, can work on several tasks at once, and raises decisions it needs from you. It runs locally in your terminal and talks directly to model APIs, without a backend server or remote code index, and it asks permission before changing files or running commands. Highlighted uses are code onboarding with agentic search, turning issues into pull requests with GitHub, GitLab and command-line tools, and multi-hour refactors and migrations that follow imports and run your tests. It can also use command-line tools such as Git and MCP servers such as GitHub. Access comes with a Claude Pro, Max, Team or Enterprise plan, or a Claude Console account. With a Console account, it consumes API tokens at standard API pricing. A fast mode for Opus 5.5 is in research preview at $8/$40 per million tokens.
Capabilities
- Plans, writes code, runs tests and opens pull requests
- Works on several tasks at once and surfaces decisions it needs from you
- Agentic search for onboarding to a codebase without selecting context files
- Uses command-line tools such as Git and MCP servers
- Runs locally and asks permission before changing files or running commands
- Extensions for VS Code (and Cursor) and JetBrains IDEs, plus Slack
When to use it
- You want to delegate a bug fix, tests or a multi-day migration and review the result
- You need to get oriented in a large unfamiliar codebase
- You prefer an agent that runs in your own terminal alongside your existing IDE
