AWS DevOps Agent adds release management capabilities to assess code changes before production (preview) | Amazon Web Services (opens in new tab)
AWS DevOps Agent’s new preview release-management features extend its role from post-deployment incident response to pre-production review and testing. It evaluates code changes against production requirements, organizational standards, dependency risks, and access-control practices, then performs targeted tests in isolated or production-like environments. The goal is to help teams safely handle the growing volume of AI-generated code without sacrificing review quality or delivery speed. ## Release Readiness Reviews - Reviews changes for: - Production and dependency risks - Cross-repository impacts - AWS access-control changes and Well-Architected best practices - Compliance with organization-specific standards - Teams can provide standards in plain English, such as: - Encryption and network-access rules - Logging and observability requirements - Sensitive-data classification practices - Without custom instructions, the agent applies general best practices. - It runs lightweight user-journey tests in an AWS-managed isolated environment to confirm that the software builds, runs, and passes basic functional checks. - Findings are available in: - The AWS DevOps Agent console - GitHub or GitLab pull-request comments - IDE workflows through the Kiro power or Claude Code plugin ## Autonomous Release Testing - Generates test plans based on the specific code change rather than relying only on static test suites. - Tests web and API applications in customer-provisioned, production-like environments before merging. - Covers: - Functional correctness - Behavioral regressions - Integration scenarios - Produces structured artifacts for every run, including metrics, logs, traces, and execution summaries. ## Configuring and Running Reviews - At least one GitHub or GitLab repository must be connected to an AWS DevOps Agent Space. - The agent indexes connected code and builds a knowledge graph of cloud and cross-repository dependencies. - Reviews can be triggered by: - Submitting a pull request - Starting an on-demand chat request, such as “Perform a production risk analysis on my repository branch” - The target can be specified using a branch name, pull-request number, or commit SHA. - Reviews can also be initiated from supported development environments. ## Reviewing Results - The **Changes** section lists review executions and supports filtering by category or status. - The **Timeline** records the agent’s tools, consulted dependencies, observations, and timestamped reasoning steps. - The **Report** includes: - Recommended action: **BLOCK**, **Proceed with Caution**, or **Safe to Release** - Number of critical issues - Commit revision and changed-file count - Evidence supporting the recommendation - Severity-ranked findings - Actionable remediation steps - A file-by-file summary of modifications - Developers can ask follow-up questions about affected downstream consumers, impacted files and line numbers, and recommended fixes. AWS DevOps Agent’s preview release-management capabilities provide an automated layer of change analysis and targeted testing before production. Teams should configure organization-specific instructions, connect their repositories, and use the generated reports and test artifacts as an additional safety gate for AI-assisted development.