What is Claude Code?

What it is

Claude Code works inside a software project. It can read the files in a codebase, propose changes, edit the files, and run commands and tests to see whether the change works. It is available in a terminal, a desktop app, on the web and inside code editors. It is an example of agentic AI, and it is powered by Claude models.

The user decides what the tool may read, change or run, and can approve each step or allow routine ones. Its changes are ordinary edits to files, so they can be reviewed and, when the project is in Git, undone.

Why it matters to a business

• Faster routine work: small fixes, tests and repetitive edits take less of an engineer's time.
• Permission settings: the tool can only do what the user permits, so check the settings and the account it runs under.
• Wrong code: a tool that writes code can write wrong code. A person should read and test every change before release.
• Data exposure: the code and files it reads are sent to an outside service, so keep secrets such as passwords and keys out of the project.

How KawaConnect works with Claude Code

Our engineers use Claude Code, and other agents such as Codex, in software work. The code lives in version control, and an engineer reads each change the agent makes before it is released. We work this way on the custom tools and integrations we build under software and AI and automation. We do not claim any partnership with Anthropic.

Learn more

• Claude Code product pageThe maker's description of the tool, for owners deciding whether it fits their team.
• Claude Code documentationThe official overview, for the engineer who will set it up and manage permissions.