Agent Skills

Skills are portable instruction packages that follow the open Agent Skills standard. They bundle instructions, scripts, and templates into reusable packages that Kiro can activate when relevant to your task.

Kiro supports the Agent Skills standard, so you can import skills from the community or other compatible AI tools, and share your own skills across the ecosystem.

AI agents are increasingly capable, but they often lack the specific context needed for real work. Without knowledge of your team's deployment process, your company's code review standards, or your project's data analysis pipeline, agents guess and iterate - just like you would when learning something new.

Loading all this context upfront isn't practical either. Too much information overwhelms the agent, slowing responses and reducing quality.

Skills solve this with progressive disclosure:

1. Discovery - At startup, Kiro loads only the name and description of each skill
2. Activation - When your request matches a skill's description, Kiro loads the full instructions
3. Execution - Kiro follows the instructions, loading scripts or reference files only as needed

This keeps context focused while giving Kiro access to extensive specialized knowledge on demand.

Skills can activate in two ways:

• Automatically - Kiro matches your request against skill descriptions and loads the relevant skill
• As slash commands - Type / followed by the skill name to invoke it directly

Skills can be created with a workspace scope or a global scope.

When skills share the same name, workspace skills take priority over global skills. This allows you to define global skills that generally apply to all your workspaces, while preserving the ability to override them for specific projects.

{
  "name": "my-agent",
  "resources": [
    "skill://.kiro/skills/*/SKILL.md",
    "skill://~/.kiro/skills/*/SKILL.md"
  ]
}

The skill:// URI scheme supports specific paths, glob patterns, and home directory expansion.

A skill is a folder containing a SKILL.md file:

my-skill/
├── SKILL.md           # Required
├── scripts/           # Optional executable code
├── references/        # Optional documentation
└── assets/            # Optional templates

The file starts with YAML frontmatter followed by markdown instructions:

---
name: pr-review
description: Review pull requests for code quality, security issues, and test coverage. Use when reviewing PRs or preparing code for review.
---

## Review checklist

When reviewing a pull request:

1. Check for vulnerabilities, injection risks, exposed secrets
2. Verify edge cases and failure modes are handled
3. Confirm new code has appropriate tests
4. Ensure variables and functions have clear names

## Common issues to flag

- Hardcoded credentials or API keys
- Missing input validation
- Unhandled promise rejections
- Console.log statements left in production code

See the full specification for detailed field constraints.

For extensive documentation, use a references/ folder:

aws-deployment/
├── SKILL.md
└── references/
    ├── ecs-guide.md
    └── troubleshooting.md

Reference the files in your SKILL.md:

For ECS deployments, follow the guide in `references/ecs-guide.md`.

Kiro loads reference files only when the instructions direct it to.

A CDK deployment skill:

cdk-deploy/
├── SKILL.md
└── references/
    └── stack-patterns.md

SKILL.md:

---
name: cdk-deploy
description: Deploy AWS CDK stacks with best practices. Use when deploying infrastructure, running cdk deploy, or troubleshooting CDK issues.
---

## Deployment workflow

1. Run `cdk synth` to validate templates before deploying
2. Use `cdk diff` to preview what will change
3. Run `cdk deploy` and review IAM changes

## Pre-deployment checks

- Verify AWS credentials are configured for the target account
- Check that the CDK version matches the project's requirements
- Review `references/stack-patterns.md` for environment-specific patterns

## Rollback procedure

If deployment fails:
1. Check CloudFormation console for the specific error
2. Run `cdk destroy` only if the stack is in a failed state
3. Fix the issue and redeploy

Usage:

> Deploy my CDK stack to staging

I'll follow the deployment workflow. First, let me synthesize the templates...

Skills are portable packages following an open standard. They load on-demand and can include scripts. Use for reusable workflows you want to share or import from others.

Steering is Kiro-specific context that shapes agent behavior. It supports always, auto, fileMatch, and manual modes. Use for project standards and conventions.

Powers bundle MCP tools with knowledge and workflows. They activate dynamically based on context. Use for integrations where you need both tools and guidance.

Write precise descriptions - The description determines when Kiro activates the skill. Include specific keywords and actions that match how you'd phrase requests:

• Good: Review pull requests for security vulnerabilities and test coverage. Use when reviewing PRs or preparing code for review.
• Vague: Helps with code review

Keep SKILL.md focused - Put detailed reference material in references/ files. Kiro loads the full SKILL.md on activation, so keep it actionable.

Use scripts for deterministic tasks - Validation, file generation, and API calls work better as scripts than LLM-generated code.

Choose the right scope - Global for personal workflows you use everywhere (your review checklist). Workspace for team procedures and project-specific conventions.

Version control workspace skills - Commit .kiro/skills/ to your repository so the team shares the same workflows.

• Steering - Project-specific context and standards
• Powers - MCP integrations with bundled knowledge
• Custom agents - Agent configuration and resources
• Agent Skills specification - Full format details