AWS Kiro Masterclass: Spec-Driven Agentic AI Development

Explore the evolution of AI-powered software development on AWS, from Code Whisperer to Kiro, highlighting spec-driven development, requirement-driven design, and interactive refinement leading to code generation.

Introduce spec driven development with Kiro, turning prompts into requirements, design, code, and tests via markdown specs. Compare wipe coding and spec driven, emphasizing production-grade benefits, documentation trails, and traceability.

Explore Kiro architecture and its 8 capabilities, showing how a developer prompt drives plan, design, and code tasks through steering files, internal tools, and MCP tools.

Download and install kiro from kiro.dev, select your OS, install the IDE; authenticate via Google for free 50 credits, or via AWS IAM Identity Center, then access IDE and chat.

Set up aws iam identity center, create and invite users, and sign in via a Kiro dashboard in us east 1 or frankfurt, selecting a Kiro plan and monitoring credits.

learn to craft spec-driven requirements with Kiro by turning a vehicle diagnostic feature into a requirements.md with user stories, acceptance criteria, and glossary.

Kiro uses the Model Context Protocol (MCP) to integrate with enterprise tools like Jira and Confluence, turning natural language requests into tool calls to push user stories and design docs.

Provide project context to kiro by steering files, guiding architecture, tech stack, and structure with product.md, tech.md, and structure.md to enforce whitelisted AWS services and naming conventions from the start.

Implement steering docs in Qiro hands-on, transition from TypeScript and Postgres to Python with DynamoDB, API Gateway, Bedrock, and Lambda, and update the design and task docs.

Execute spec-driven tasks to build FleetMate end-to-end, provisioning and deploying the AWS stack with CloudFormation and SAM, configuring IAM, Cognito, and Lambda services for a working UI.

Learn how Kiro hooks automate repetitive development tasks by triggering agent prompts or shell commands on IDE events to update documentation and tests, accelerating delivery and improving quality.

Learn to create and configure agent hooks in Kiro to automate tasks like code documentation, cleanup, and security pre-commit scans, with event triggers and actions.

Explore Quiro powers, bundles of documentation, steering files, and MCP server configurations that enable context-aware loading by matching prompts to the right power.

Demonstrates hands-on implementation of Kiro powers, focusing on AWS observability, installing and configuring five MCP servers, and enabling cloud watch logs, metrics, alarms, and application signals.

Explore Kiro’s large language model options and AWS usage guidance, including auto, opus, haiku, and sonnet, with cost, context windows, and task suitability.

Optimize enterprise app development with QIRO by balancing quality, cost, and delivery through right foundation model selection and essential context like design documents and task.md.

Discover MCP best practices for Kiro: replace huge prompts with on-demand external context from tools like Jira and Confluence, optimize tool descriptions, and enforce secure, selective tool usage.

Harness AI agents powered by large language models to solve complex tasks through planning, tools, memory, guardrails, and communication, enabled by retrieval augmented generation for vacation planning data access.

Learn how the model context protocol standardizes data sharing between large language models and data sources, detailing host, client, server, transport, tools, resources, and prompts.