Claude Code
Claude Code is an agentic coding tool that lives in your terminal: it reads your repository, plans changes, edits files, runs tests, and explains what it did. It is the most-cited example of what “agent” means in practice.
How it works: it is a CLI that drives Claude with tool access (read/write files, run shell commands, search). You give it a task in plain language; it iterates until the task is done, asking when it needs a decision.
Pricing: included in Claude Pro/Max subscriptions; API billing available.
Best for: developers who want an agent inside their existing workflow — refactors, test writing, migration work — without leaving the terminal.
Details
| Type | agent |
|---|---|
| Pricing | freemium |
| Local model support | Yes |
| Cloud model support | Yes |
Pros
- Deep, multi-file agentic workflow: plans, edits, runs tests, iterates
- Lives in the terminal — fits into any existing dev workflow
- Can point at local models via Ollama
Cons
- Requires a paid Claude subscription or API budget for serious use
- Terminal-only — no visual diff/IDE experience