What you'll learn:
- Turn a single prompt into requirements, design and task documents using Kiro's spec-driven development workflow
- Write steering files that force every Kiro response to follow your project's coding standards and conventions
- Build custom Kiro skills from SKILLmd, including skills that run your own Node.js scripts to save time and tokens
- Configure MCP servers in mcp.json and connect Kiro to live data, including the official AWS MCP server
- Use Kiro Powers to load only the tools a task actually needs instead of bloating your context with every MCP tool
- Create custom agents with scoped tool access, per-agent model selection and pre-approved permissions
- Automate your workflow with agent hooks that fire on file changes, tool calls and prompt submissions
- Run Kiro from the terminal with Kiro CLI, using slash commands, JSON agents and parallel sessions
AWS Kiro is Amazon's agentic IDE — a VS Code fork built around spec-driven development. Instead of turning your prompt straight into code, Kiro writes a requirements document, a technical design and an ordered task list that you review and edit before anything is built.
This course covers every major Kiro feature, hands-on: spec-driven development with requirements md, design md and tasks md; EARS acceptance criteria; steering files that enforce your coding standards; custom skills built from SKILLmd; MCP servers configured in mcp json, including the official AWS MCP server; Kiro Powers for controlling context bloat; custom agents with scoped tools and permissions; agent hooks; and the Kiro CLI.
You start by installing Kiro IDE and signing in with an AWS Builder ID, and finish running Kiro from the terminal in parallel CLI sessions. Each concept gets a short explanation, then we build it live: a steering file that forces AI-generated constants into your naming convention, a skill that runs a Nodejs script, a hook that blocks a tool call on a keyword, an agent restricted to commands you pre-approve.
All code, generated spec documents and prompts are attached as downloadable resources. Optional lectures on VS Code basics and agentic AI fundamentals are marked, so you can skip what you already know.
What you'll learn
Spec-driven development — turn a prompt into requirements, design and task documents, write acceptance criteria in EARS syntax, and let Kiro build a dependency graph that runs tasks in parallel waves. You'll also learn when to use plan mode instead.
Configuring Kiro — generate project steering documentation, then write custom steering with conditional inclusion that activates only on matching files. Build skills from scratch, including one that executes your own script.
MCP and Kiro Powers — what the Model Context Protocol solves, how to configure servers at workspace or user scope, and how to install the official AWS MCP server so Kiro answers from current AWS documentation. Then use Powers to stop stacked MCP servers from destroying your context window.
Agents, hooks and the CLI — scope agents to specific tools, models and permissions; automate with hooks that fire on file changes, tool calls and prompt submissions; and run everything from the terminal with Kiro CLI.
An honest note on scope
This is a deep course on Kiro, not an app-building bootcamp — the hands-on work uses a small Vite and React project as a vehicle for learning the tool, so you can apply it to any stack. You'll also see Kiro's rough edges: where a feature was buggy at recording time, I say so and show the workaround.
