Claude Code

Claude Code, from Anthropic. An agent that reads your project, runs commands and makes changes from the terminal or IDE.

What it does well

• Multi-step programming tasks end to end
• Data-cleaning scripts you can rerun
• Understanding an unfamiliar repository

What it does poorly

• Running it with permissions you don't understand
• Tasks you can't verify

What makes it different

It works agentically: plans, edits files, runs tests and iterates, while you approve what it does.

Quick start

1. Install Claude Code and open a terminal in a practice repository.
2. Ask it to explain the project structure.
3. Give it one small, testable task and review the diff.
4. Add a CLAUDE.md with your project's conventions.

Strengths and limits

• Strength: Handles long, multi-step tasks
• Strength: Works in terminal, IDE and web
• Strength: Extensible with skills and MCP
• Limit: Needs a paid plan or API credits
• Limit: Requires comfort with the command line

Privacy

It can read files in the folders you open. Don't point it at sensitive data.

Academic integrity

Check course rules on agentic coding tools.