
Claude Code: Anthropic's agentic coding tool
In short: Claude Code is Anthropic's agentic coding tool for the terminal, IDE, desktop and web: what it is, how it works and who it suits.
Claude Code is Anthropic's agentic coding tool. It runs in your terminal, in IDEs such as VS Code and JetBrains, in the Claude desktop app and on the web, and it can read your codebase, edit files, run commands and tests, and work with Git. Unlike autocomplete, you give it a task in plain language and it works through the steps, asking permission before risky actions by default.
Claude Code in plain words
A capable developer who sits in your terminal, reads the project before touching it, and asks before running anything that matters.
How it works
Claude Code is an agent harness around Claude models. On each turn it gathers context (files, your CLAUDE.md instructions, command output), decides an action, asks for permission according to your settings, runs it and reads the result. It can be extended with MCP servers, hooks, skills, slash commands and subagents, and it can run non-interactively for scripts and CI.
Example
cd my-project
claude
> the checkout test fails on empty carts, find out why and fix it, then run the tests
Starting a session this way follows the quickstart. It will search the code, read the failing test, propose an edit, apply it after approval and rerun the suite.
Where you'll meet it
- Developers' terminals and IDEs on macOS, Linux and Windows
- Automated pipelines through headless mode and GitHub integrations
- Team setups with shared
CLAUDE.md, hooks and MCP servers - Paid Claude plans and API-billed usage, depending on how it is set up
What it changes for you
For individual developers, the skill shifts to specifying tasks clearly and reviewing diffs well. For tech leads, repository conventions written in CLAUDE.md become as important as the README. For organisations, cost, permission policy and code review practice need to be decided before broad rollout.
Claude Code vs Claude on claude.ai
Claude on claude.ai is a chat assistant. Claude Code is an agent that works directly in your project: it reads files, runs commands and makes changes.
Common mistake
My view: starting with a huge vague task. Ask for a plan first, then one change at a time, with tests run after each. Claude Code is strongest when the loop is short and verifiable.
Related terms
AGENTS.md and CLAUDE.md · Claude Code hooks · Agent skills · Permission modes · Plan mode · OpenCode
From the archive: A Claude Code course needs a personal workflow map
Keep the loop short: plan, one change, run the tests, review the diff.


