Tool Use
Claude Code can act because it has tools — Read, Write, Bash, and more — and it picks the right one automatically.
What separates Claude Code from a plain chatbot is tool access: built-in capabilities that let it interact with your actual filesystem and shell. The three you'll notice most are Read (look at files), Write/Edit (create or change files), and Bash (run commands — install packages, run tests, move files). You never choose the tool yourself; you describe the goal and Claude Code selects the tool.
What's changed since the video: the tool roster has grown well beyond those three. Depending on context you'll also see a Task tool for delegating to subagents, a Monitor tool for watching background processes and log output, an AskUserQuestion tool for structured clarifying questions, and tools specific to worktrees, browser automation and MCP-connected services.
Real-life use cases
One instruction, several tools chained
“Find why the build is failing and fix it” might trigger Bash (run the build, read the error), Read (open the failing file), Edit (apply the fix), and Bash again (re-run the build to confirm) — all without you naming a single tool.
Safe read-only exploration
Before allowing any changes, a cautious user pre-approves only the Read tool so Claude Code can explore and explain an unfamiliar codebase without being able to modify anything yet.
Source
- Source video: “Claude Code Explained” — 27 concepts in 27 minutes (YouTube)https://www.youtube.com/watch?v=ZlDnsf_DOzg
- Tools referencehttps://code.claude.com/docs/en/tools-reference