Updated November 2026

Spec-Driven Agentic Coding with Kiro: Specs, Steering, Hooks, and the CLI

Class Duration

14 hours of live training delivered over 2-3 days to accommodate your scheduling needs.

Student Prerequisites

Professional software development experience in at least one language
Working knowledge of Git and GitHub or GitLab
Comfort with a VS Code-style editor and the command line
A Kiro account on an individual plan or an organization subscription
Experience with Amazon Q Developer is helpful for the migration topics
Familiarity with spec-driven development is helpful; see Spec-Driven Development with AI Coding Agents

Target Audience

Software engineers, tech leads, and platform teams adopting Kiro as their agentic coding environment. AWS teams moving off Amazon Q Developer, and administrators who will roll Kiro out across an organization, will also benefit.

Description

Kiro is an agentic development environment from AWS, with one agent harness behind its IDE, CLI, web, and mobile apps. This course teaches Kiro's spec workflow: EARS requirements, a design document, and a task list that agents carry out. Students add steering, hooks, powers, and skills, then run Kiro CLI in the terminal and in CI. The course also covers permissions, credits, and governance. It ends with a migration plan for Amazon Q Developer, whose IDE plugins and paid subscriptions reach end of support on April 30, 2027.

Learning Outcomes

Set up Kiro IDE and Kiro CLI with individual or organization sign-in.
Choose a session type for the scope and risk of the work.
Write feature specs with EARS requirements, designs, and task lists.
Fix defects with bugfix specs that preserve unchanged behavior.
Check implementations against requirements with property-based tests.
Encode project conventions with steering files and AGENTS.md.
Automate quality checks with hooks.
Extend Kiro with powers, MCP servers, skills, and custom agents.
Run Kiro CLI headless in CI and in cloud sessions.
Control agent autonomy with permission modes and the sandbox.
Manage credits, models, and enterprise governance.
Plan a team migration from Amazon Q Developer.

Training Materials

Comprehensive courseware is distributed online at the start of class. All students receive a downloadable MP4 recording of the training.

Software Requirements

Kiro IDE and Kiro CLI installed, a Kiro account (AWS Builder ID, GitHub, Google, or an organization sign-in) on a paid plan or organization subscription for the headless and cloud session topics, Git, and a GitHub or GitLab account with a repository.

Training Topics

Kiro Foundations and Setup

Kiro IDE, CLI, Web, and Mobile
One shared agent harness
Kiro IDE on Code OSS
Importing VS Code settings
Open VSX extensions
Signing in with kiro-cli login
Builder ID and IAM Identity Center
External identity providers
Project and user configuration
Configuration Sync and workspace trust

Default, Plan, and Spec Sessions

The session type picker
Default, Spec, and Plan sessions
Bug Fix and Quick Spec sessions
Matching the session to the work
Plan sessions with /plan
Context providers such as #codebase
Checkpoints and Revert
/rewind and /tangent in the CLI
Compaction for long sessions

Writing and Refining Feature Specs

Spec files in .kiro/specs/
requirements.md, design.md, and tasks.md
Requirements-first and design-first workflows
Quick Spec for well-understood features
EARS acceptance criteria
Finding gaps with Analyze Requirements
Architecture and error handling in designs
Iterating before any code

Executing Tasks and Fixing Bugs with Specs

Running tasks one at a time
Parallel task waves
Tracking task status
Keeping specs in sync
Bugfix specs and bugfix.md
Current, expected, and unchanged behavior
Root cause analysis
Property-based tests from requirements
Specs in the terminal with /spec

Steering for Project Context and Conventions

Workspace and global steering
product.md, tech.md, and structure.md
Inclusion modes
fileMatch patterns
AGENTS.md support
File references in steering
Excluding files with .kiroignore
Sharing team steering

Hooks for Event-Driven Automation

JSON hook files in .kiro/hooks/
Global hooks for Kiro CLI
Creating hooks in natural language
Prompt, agent, and session triggers
Tool use and file triggers
Task and compaction triggers
command and agent hook actions
Blocking risky tool calls
Lint and test on save

Powers, MCP, Skills, and Custom Agents

MCP servers in mcp.json
Remote servers and OAuth
Powers that load on demand
The kiro.dev/powers catalog
Agent Skills in .kiro/skills/
Custom agents in .kiro/agents/
Agent prompts, tools, and permissions
Sub-agents for parallel work

Memory and Workflows

Global and repository memories
Memory access levels
Reviewing and deleting memories
Multi-agent workflows in .kiro/workflows/
Built-in investigate and publish-pr workflows
Running workflows with /workflow run
Fresh context per step

Kiro CLI in the Terminal and in CI

The V3 and V2 agent harnesses
Upgrading agents with /upgrade-agent
Core slash commands
Spawning sessions with /spawn
Headless mode with --no-interactive
KIRO_API_KEY and --trust-tools
JSON Lines output
Exit codes and --require-mcp-startup
Pull request review in GitHub Actions
Kiro in JetBrains and Zed over ACP

Supervision, Sandboxing, and Cloud Sessions

Supervised and Accept file changes modes
Approve for me mode
Capability rules in permissions.yaml
Workspace trust
When to use --trust-all-tools
The local command sandbox
Sandboxing stdio MCP servers
Cloud sessions in Agent Focus Mode
Cloud sessions with kiro-cli chat --cloud
Kiro Web and Kiro Mobile sessions
PR feedback with /kiro fix

Credits, Models, and Enterprise Administration

How credits are consumed
Per-model credit multipliers
Claude, GPT, and open-weight models
Tracking usage with /usage
Kiro profiles and the console
Organization identity and credit pooling
Model and MCP governance controls
Enterprise Management policies
Usage dashboards and prompt logs
AWS CloudTrail and OpenTelemetry

Migrating Teams from Amazon Q Developer

What reaches end of support
Moving VS Code users to Kiro IDE
JetBrains users and Kiro over ACP
Visual Studio and Eclipse users
Q Developer CLI to Kiro CLI
Existing .amazonq/ rules
Converting rules to steering
AWS Transform for Java and .NET
AI-DLC workflows in Kiro
Rollout with governance first