Claude Código 101

Resumen

AI coding agents are changing what it means to write software. Tasks that used to take an afternoon — tracing a bug across a large codebase, scaffolding a new service, reviewing a stack of pull requests — can now happen in a single focused session, with an agent that reads your code, runs your commands, and edits files alongside you.

But getting real value from an agent requires more than installing it and typing a request. It requires understanding how the agent thinks, what context it has access to, and how to steer it when it heads in the wrong direction.

This course teaches developers how to use Claude Code effectively, whether you're new to software engineering or an experienced engineer who hasn't yet worked with AI coding agents. We start from first principles — what an agentic loop actually is, how the context window shapes what Claude can see, how tools and permissions determine what it can do — so that the techniques later in the course make sense rather than feeling like a list of tricks to memorize.

You'll learn how to install Claude Code across multiple environments (terminal, VS Code, JetBrains, Claude Desktop, and the web), and how to write prompts that get good results on the first try using approval mode, auto-accept, and Plan Mode. The core of the course is the Explore → Plan → Code → Commit workflow:

a repeatable rhythm for breaking down a task, letting Claude propose an approach, reviewing the work as it happens, and landing it cleanly.

We also cover code review with Claude Code and the context-management commands (/compact, /clear, /context) that keep long sessions productive. The final section is about making Claude Code your own.

You'll write a CLAUDE.md file so Claude remembers your project's conventions across sessions, build custom subagents and skills for tasks you repeat often, connect external systems through MCP servers, and write hooks that add deterministic guardrails around what Claude is allowed to do. By the end, you'll have a setup tailored to how you actually work — not a generic install.

Recommended prerequisites Basic familiarity with a code editor and the command line. You'll also need a Claude account (Pro, Max, or Enterprise) or an API key.

No prior experience with AI tools is assumed. Who this is for New developers entering software engineering who want AI-assisted workflows from the start, and experienced engineers curious about coding agents but who haven't taken the plunge yet.

If you've tried a coding assistant before and found the results underwhelming, this course is designed to show you what changes when you work with an agent rather than against it.

Programa

• ¿Qué es Claude Code?
Antes de escribir tu primer prompt, es útil saber qué sucede realmente cuando se ejecuta Claude Code. Esta sección explica qué diferencia a un agente de codificación de inteligencia artificial de un asistente basado en chat, luego recorre el ciclo agente: reunir contexto, actuar, verificar resultados, y las herramientas y permisos que rigen cada paso. Saldrás con un modelo mental que hará que todo en el resto del curso tenga sentido.
• Tu primer prompt
Haz que Claude Code funcione donde ya trabajas: en el terminal, dentro de VS Code o JetBrains, en Claude Desktop, o en la web. Luego, escribe tu primer prompt y observa cómo el modo de aprobación, la aceptación automática y el Modo Plan cambian la forma en que Claude responde, para que puedas elegir el nivel adecuado de supervisión para la tarea que tienes frente a ti.
• Flujos de trabajo diarios
Aquí es donde Claude Code se convierte en parte de tu forma de trabajar. Aprende el ritmo Explorar → Planificar → Codificar → Confirmar para abordar tareas reales, usa los comandos de gestión de contexto para mantener las sesiones largas rápidas y enfocadas, y pon a Claude a trabajar revisando código —el tuyo y el de tus compañeros— antes de que se integre.
• Personalizando Claude Code
Tal como viene, Claude Code es de propósito general. Esta sección te muestra cómo hacerlo tuyo: escribe un archivo CLAUDE.md para que recuerde las convenciones de tu proyecto, construye subagentes y habilidades para los flujos de trabajo que repites, conecta sistemas externos a través de servidores MCP, y agrega ganchos para límites deterministas. Al final, tendrás una configuración ajustada a la forma en que tu equipo realmente trabaja.

---

Materias

Artificial Intelligence