Claude Code

Claude Code is Anthropic’s coding agent, and it is different from both completion tools and AI editors: it runs in your terminal and decides for itself which files to read, which commands to run and what to change, then shows you the result. You describe the task; it does the work, rather than handing you a snippet to paste.

It fits tasks that span several files and require understanding the current state first: tracking down a bug, adding tests to existing behaviour, migrating a batch of code to a new convention. It reads how things are done before changing them. Because it lives in the terminal it is not tied to any editor, and it can also connect to IDEs such as VS Code and JetBrains for reviewing changes.

Worth being deliberate about: it can execute commands. It asks before each step by default, but understand what it is capable of before loosening that — especially in a directory with uncommitted work or production configuration.

Key features

• Autonomous code work: searches and reads the relevant files, plans a change, edits across files and shows you the diff.
• Command execution: runs tests, builds and Git commands, then adjusts based on the output — an edit, run, fix loop.
• Permission controls: asks before sensitive actions by default; you can widen or narrow which commands are allowed.
• Project memory: a CLAUDE.md file in the repo holds conventions and common commands, and it reads it on every start.
• Extensibility: connect external tools and data through MCP, and add custom commands and hooks.

How to use

1. Get a Claude subscription or an Anthropic API account.
2. Install Claude Code in your terminal following the official docs, and sign in on first run.
3. Start it inside your project directory and describe a task in plain language, such as “find out why this test fails and fix it”.
4. Review each proposed command and change before approving it; when you are happy, have it commit for you.

Best for

• Debugging: it reads logs and code and reproduces the issue itself, instead of you pasting errors back and forth.
• Adding tests to existing code: it works out current behaviour, then writes tests and gets them passing.
• Bulk migrations and refactors: convert a batch of files to a new convention and verify with the test suite.
• Developers comfortable on the command line who do not want to change editors.

Strengths and limitations

Strengths

• Genuinely agentic: it closes the loop of reading, editing and verifying.
• Editor-agnostic, so it fits any development setup.
• Deeply customisable through project files, MCP and hooks.

Limitations

• It can run commands; loosen permissions carefully and work under version control.
• Heavy use is expensive, and pay-as-you-go API costs are hard to predict.
• Terminal-only interaction is a barrier for people who avoid the command line.
• It shares Claude’s account system, so mainland China is not a supported region.