Claude Code Foundations – Your Daily Dev Workflow By the end of this lesson, students should be able to: Claude Code is not a chat window. Claude Code is a terminal-based AI coding assistant. It runs in your existing terminal, operates in the context of your local file system, can read and edit your actual files, run shell commands, and reason about your real codebase – not a pasted excerpt you give it. This architecture produces a fundamentally different kind of AI interaction than copy-pasting code into a browser chat window. Three modes. Claude Code operates in three modes (verify current mode names and behavior at docs.anthropic.com): Chat mode ( Code mode (formerly Bash/code execution): Claude can read files, edit code, run commands, and operate tools. The core working mode for actual development tasks. Claude acts on your files. Cowork/Agent mode: Extended autonomous operation for multi-step tasks. Claude can complete a sequence of actions – reading, editing, testing, iterating – with defined check-in points. Most powerful for complex tasks with clear goals. CLAUDE.md – persistent project context. CLAUDE.md is a markdown file you place at the root of a project directory. Claude Code reads it automatically at the start of every session in that project. It is your standing context document – covering the project's architecture, conventions, current focus, and any preferences Claude should apply consistently. Without CLAUDE.md, Claude Code starts each session without knowing anything about your project. With a good CLAUDE.md, Claude starts informed: the tech stack, the naming conventions, the current feature in progress, the things to avoid. This is the single highest-leverage configuration investment for a regular Claude Code user. What Claude Code can and cannot do. Can do: read and edit files, run shell commands, execute test suites, navigate directory structures, use tools like web search (when enabled), reason about multi-file codebases. Cannot do: access your running application at runtime, observe live browser state, connect to external APIs without an explicit tool or plugin, remember between sessions without CLAUDE.md or Projects. A backend developer installs Claude Code and opens it in a new terminal in her project directory. She types /chat and asks a question about her codebase architecture. Claude responds without reading anything – it knows nothing about her project yet. She creates a CLAUDE.md file with five lines: tech stack, active feature, test command, style guide location, and the pattern to use for database queries. On her next session, Claude's first response already reflects the project context she needs. The CLAUDE.md took twelve minutes to write and eliminated the context re-establishment that was consuming the first three to five minutes of every session. Claude Code in Code mode has write access to your file system. It can edit and delete files. Before running Claude Code in agent mode on a production codebase or in a directory with irreplaceable files, understand the scope of its permissions. Use version control (git) so any unwanted changes can be reverted. For sensitive or production systems, verify the current permission model and recommended settings at docs.anthropic.com. Log in and enroll to access lesson quizzes.
Lesson 1: What Claude Code Is and How It Works
Lesson Objectives
Lesson Content
/chat or default): Conversational AI assistance. Discuss code, get explanations, brainstorm. Claude does not take actions on your file system in this mode – it only responds in text.Practical Example
Safety Notes