GitLab Duo concepts
Here are some concepts and terms to know when using
GitLab Duo with Amazon Q
Topics
Setting up GitLab Duo with Amazon Q
Before you can use Amazon Q artificial intelligence (AI) capabilities in GitLab Duo, you need
to complete the prerequisites and create AWS resources. For more information, see
Set up GitLab Duo with
Amazon Q
Onboarding with AWS resources and permission policies
As part of the GitLab Duo onboarding process, you need to create an Amazon Q Developer profile through the
Amazon Q Developer console
When the new IAM role is created, the required trust policy with the necessary permissions is also created. A role trust policy is a required resource-based policy that is attached to a role in IAM.
You need to add a permissions policy, which grants ability to connect with Amazon Q and utilize the features in the GitLab Duo with Amazon Q integration. The policy must be added when creating the IAM role. To learn more about the permissions provided by the permissions policy, see GitLabDuoWithAmazonQPermissionsPolicy.
Alternatively, you can create an inline policy and add the required permissions. You can choose to create an inline policy if you want to custom access control. For more information, see Managed policies and inline policies and Policies and permissions in AWS Identity and Access Management in the IAM User Guide.
Trust policy
Permissions policy
Optionally, you can also use customer managed keys (CMK) to encrypt your resources
if you want full control over the lifecycle and usage of your key. The kms:ViaService
condition key to limit who can use CMK for encrypting and decrypting content. For more
information, see Manage access to Amazon Q Developer for
third-party integration.
GitLab quick actions
When invoked, quick actions perform tasks for you in GitLab issues and merge
requests. To learn how to invoke quick actions in GitLab, see the
GitLab
documentation
Merge request generation and iteration
-
/q dev– Allows you to go from a high-level idea captured in a GitLab issue to having Amazon Q generate a ready-to-review merge request with the proposed code implementation. This helps streamline the process of turning concepts into working code. The merge request is created in a new branch and Amazon Q assigns the issue creator as a merge request reviewer. You're also provided a merge request summary. For more information, see Turn an idea into a merge request.
Code review
-
/q review– Allows you to initiate a merge request review in GitLab Duo with Amazon Q. An automatic code review is initiated for new merge requests. As a GitLab administrator, you can also configure Amazon Q to turn off automatic reviews. Automated code reviews identify and fix potential issues as Amazon Q generates and suggests code fixes to your merge request. They provide quality checks, analyzing for issues, logical errors, anti-patterns, code duplication, and more.Amazon Q gives you code analysis with comments, with each comment providing a separate finding. This quick action is available for all languages. Automatic code reviews are initiated when you open new merge requests or reopen previously closed ones. However, automatic code reviews won't be triggered by subsequent commits made within an existing merge request. You can manually trigger a code review by using the
/q reviewquick action.You can configure code reviews to run automatically on every new merge request within your GitLab instance or group. For more information, see Review a merge request
.
Chat session in web UI and IDEs
-
GitLab Duo Chat and Code Suggestions works with Amazon Q to provide support for CI/CD configuration, error explanations, and addressing questions. You can use slash commands in a chat session to invoke the GitLab Duo with Amazon Q chat capabilities. For more information, see Ask GitLab Duo Chat
.