Setup & First Run

Get started with Kiro Web (Preview) by signing in and connecting a repository provider. Kiro Web works with both GitHub and GitLab. Setup steps differ depending on whether you use social login or AWS Identity Center.

1. Navigate to app.kiro.dev and sign in with your AWS Builder ID
2. Ensure you have a paid Kiro subscription (Pro or higher). Kiro Web is not available on the free tier.
3. Connect a repository provider — GitHub or GitLab — to give the agent access to your repositories

1. Your administrator must enable the Kiro Web agent for your organization
2. Sign in at app.kiro.dev using your Identity Center credentials
3. Kiro Web requires a Kiro Profile — it does not work with Q Developer Profiles
4. Connect a repository provider — GitHub or GitLab — to give the agent access to your repositories

See the AWS Identity Center guide for details on shared settings and limitations.

Connect your GitHub account to give the agent access to your repositories:

1. Go to Settings and select the Agent tab
2. Under GitHub, click Connect GitHub
3. Authorize the Kiro Agent GitHub app
4. Select which repositories the agent can access

You must have write permissions on repositories for the agent to create branches and open pull requests.

How repository access works: Kiro Web shows all repositories where both conditions are met:

1. Your GitHub user has access to the repository
2. The Kiro Agent GitHub app has been installed and authorized for that repository

This means you'll see repositories from personal accounts, shared repositories, and organizations — as long as both your GitHub user and the Kiro Agent app have access.

See the GitHub integration guide for detailed setup instructions.

Kiro Web connects to GitLab using a personal access token (PAT):

1. In GitLab, select your avatar and choose Preferences, then go to Access > Personal access tokens and select Generate token
2. Select a scope, then create the token and copy it. The api scope gives Kiro full access; see the GitLab integration guide if you want to scope the token down
3. In Kiro, go to Settings and select the Agent tab
4. Under GitLab, click Connect GitLab
5. Paste your personal access token and select the GitLab instance (gitlab.com is populated by default)

Once connected, your GitLab projects are available to select when you create a session or add a repository mid-session.

See the GitLab integration guide for detailed setup instructions.

Once you've connected a repository provider, you're ready to put the agent to work.

1. Go to app.kiro.dev
2. Optionally select repositories using Select repo before starting your task

Write a clear description of what you want the agent to do. For your first task, try something straightforward like:

• "Add error handling to the login function in auth.ts"
• "Write unit tests for the UserService class"
• "Update the README with installation instructions"

Or try something that spans multiple repositories:

• "Add a new API endpoint in the backend service and update the frontend client to call it"
• "Update the shared authentication library and migrate both the web and mobile apps to use the new version"

Include any specific requirements or constraints in your description. You can always refine or add more details as the conversation progresses.

If you didn't select repositories when starting the session, you'll be prompted to choose them when you ask the agent to work on something. Only select repositories you trust, especially when mixing public and private repos, as the agent learns from and follows instructions in the repository code even if those instructions are malicious.

How the agent proceeds depends on the mode you're using:

Default mode — The agent works with you interactively. It analyzes your request, discusses the approach, and you iterate together. When you're satisfied, ask the agent to open a pull request.

Autonomous mode — Toggle Autonomous on in the chat input bar. The agent asks clarifying questions, builds a plan, executes the work with specialized sub-agents, and opens pull requests automatically.

When the agent opens a pull request:

1. Check the code changes and implementation approach
2. Leave comments with feedback or suggestions
3. The agent automatically addresses your comments and pushes updates

Comments like "always use our standard error handling" or "remember to follow our naming conventions" help the agent learn your team's patterns for future work.

• Learn more about working with the agent
• Try autonomous mode for end-to-end task execution
• Configure the sandbox
• Set up steering files to guide the agent