AWS Builder Center

Another day, another post on AWS Transform

I keep coming back to this service. Most of what I have done with it is POC work on single repos and small demo codebases, nothing at enterprise scale.

Cloud @ Sudo | 14x AWS Certified | 2x Claude Certified | 2x Oracle | 2x CNCF(KCNA , KCSA)
I keep coming back to this service. Most of what I have done with it is POC work on single repos and small demo codebases, nothing at enterprise scale. But I have been poking at it since fairly early on, and the distance between what it did a year ago and what it does now is wide enough that I wanted the whole catalog written down in one place.
My own path through it was pretty boring, honestly. I started with a Java version upgrade, because that is the obvious first thing anyone tries. Then Python. Then Node.js. Then the comprehensive codebase analysis transformation landed and I stopped thinking of Transform as an upgrade tool. Now there is continuous modernization, which went generally available on 3 August 2026, and the model shifts again: you point it at a GitHub org rather than a folder on your laptop, and it keeps scanning on a schedule.
For a sense of the growth, AWS said in June that Transform had processed 7 billion lines of code with roughly 2 million hours saved. A month earlier that figure was 4.5 billion.
So here is the full list. 23 transformations in the catalog right now, 9 generally available and 14 in early access.



Language version upgrades

All three are GA, and all three handle arbitrary source and target versions rather than fixed hops.
  •  AWS/java-version-upgrade  works with any build system, from any source JDK to any target JDK. It also covers Jakarta EE migration, database drivers, ORM frameworks and Spring ecosystem updates, so it does considerably more than bump a number in a pom file.
  •  AWS/python-version-upgrade  goes from Python 3.8 or 3.9 to 3.11, 3.12 or 3.13.
  •  AWS/nodejs-version-upgrade  goes from any Node version to any other, with full dependency modernization.
You set the target either by talking to the agent interactively or by passing an  additionalPlanContext  parameter, which matters if you are scripting it.



AWS SDK migrations

These are the ones I think get underrated. If you have anything written before about 2021, you probably have one of these sitting in your codebase.
  •  AWS/java-aws-sdk-v1-to-v2  for Maven or Gradle projects.
  •  AWS/python-boto2-to-boto3 , built from the official AWS migration documentation.
  •  AWS/nodejs-aws-sdk-v2-to-v3 , which moves you to the modular v3 architecture with TypeScript support and the middleware stack, and specifically does not touch your Node version. Useful if you want to decouple the two migrations.



GenAI model migration

 AWS/mke-genai-model-migration  is the most interesting thing in the catalog and the one nobody seems to be talking about.
It scans your codebase, finds every AI SDK and model you are using, then produces a migration plan to Amazon Bedrock with cost comparisons. It handles OpenAI, Google Gemini, the Anthropic SDK directly, and open source models through LiteLLM or Ollama. On the framework side it understands LangChain, LlamaIndex, CrewAI, LangGraph and multi-provider routing layers.
The part I like: it swaps only the model layer. If you have a LangGraph StateGraph with conditional edges, it changes the LLM binding and leaves your graph topology and routing logic alone. It also excludes any model within 90 days of end of life from its recommendations, and if a capability has no Bedrock equivalent it says so and suggests a partial migration instead of quietly breaking your app.



Analysis

  •  AWS/comprehensive-codebase-analysis  does deep static analysis and produces hierarchical, cross-referenced documentation covering architecture, business logic and technical debt. This is the one that changed how I use Transform. It generates a knowledge base rather than a diff.
  •  AWS/java-performance-optimization  reads Java Flight Recorder profiling data, finds CPU and memory hotspots and anti-patterns, then applies targeted code fixes. You have to collect the JFR data yourself first.
Worth noting that both of these produce reports and do not modify your code.



Early access

Functional, but AWS notes these get updated frequently based on feedback, so pin your expectations accordingly.
Architecture and runtime:
  •  AWS/early-access-java-x86-to-graviton  validates Arm64 compatibility, updates dependencies, detects architecture-specific code patterns and recompiles native libraries where source is available. AWS is upfront that many modern Java apps are already Arm64 compatible, so do not expect drama here. Run it in an Arm64 environment for best results.
  •  AWS/oracle-java-to-corretto  migrates from Oracle JDK to Amazon Corretto. This one does a lot: replaces  sun.* ,  com.sun.*  and  com.oracle.*  internal APIs, swaps container base images, updates CI workflow distribution fields, strips commercial JVM flags, handles Alpine to Amazon Linux package translation, and writes a licensing report with cost savings.
Frontend:
  •  AWS/early-access-angular-to-react-migration  for Angular to React.
  •  AWS/angular-version-upgrade  for older Angular to a target version, covering components, services, templates and routing.
  •  AWS/vue.js-version-upgrade  for Vue 2 to Vue 3. Minor and patch updates are explicitly out of scope.
Java frameworks:
  •  AWS/spring-boot-version-upgrade  handles build files and starters, properties, Jackson 3 packages, Spring Security 7 DSL, test infrastructure and observability dependencies.
  •  AWS/JBoss-to-Spring-Boot  moves Java EE or Jakarta EE apps off JBoss EAP and WildFly onto Spring Boot, dropping the app server dependency entirely.
  •  AWS/early-access-log4j-to-slf4j-migration  migrates Log4j 1.x or 2.x to SLF4J with Logback, handling source, Maven or Gradle dependencies, and config files. It validates by compiling, running tests and scanning for residual imports.
Platform migrations:
  •  AWS/oracle-service-bus-to-aws  reads OSB proxy, pipeline and business service XML plus BPEL process configs, and generates a deployable CDK TypeScript project using API Gateway, Lambda and Step Functions.
  •  AWS/datadog-monitors-to-cloudwatch-alarms  converts DataDog metric monitors into native CloudWatch Alarms as infrastructure as code. It outputs CloudFormation YAML, CDK TypeScript or Terraform HCL, and accepts JSON exports plus  .tf ,  .tfstate  and  .tfplan  inputs. It handles anomaly detection bands, metric math and Metrics Insights SQL.
Readiness assessments:
  •  AWS/agentic-readiness-analysis  evaluates whether your systems can safely be called by AI agents, looking at APIs, identity, state management, human-in-the-loop and observability.
  •  AWS/modernization-readiness-analysis  scans for cloud-native maturity gaps and maps them to AWS modernization pathways.
  •  AWS/portfolio-agentic-readiness-analysis  aggregates individual agentic readiness reports across a portfolio to find cross-cutting blockers and shared remediation patterns.
  •  AWS/portfolio-modernization-readiness-analysis  does the same for modernization reports, producing a consolidated roadmap with prioritized migration waves.
Those last two are the ones I would look at first if you are on the consulting side. They do not scan code at all, they aggregate other reports, which is exactly the shape of a portfolio assessment deliverable.



Three things that are easy to get wrong

Transform is three separate things. The service itself covers Windows, mainframe COBOL and PL/I, VMware and AI workload migrations, and none of those appear in this catalog. Transform custom is the catalog above. Continuous modernization is a third layer with its own six analysis types ( rapid-techdebt-analysis ,  tech-debt-comprehensive ,  security ,  agentic-readiness ,  modernization-readiness ,  custom ), where the  custom  type is what lets you run a catalog transformation as a scheduled analysis. Most confusion I have seen about Transform comes from mixing these up.
Do not trust any published list, including this one. AWS describes the catalog as continuously growing, and names have already changed. An AWS blog from April 2026 documents  AWS/early-access-comprehensive-codebase-analysis , which has since dropped the prefix. Copy that command today and you get an error. Run  atx custom def list  against your own registry and region instead.



If you want to try it

The catalog is the canonical reference, and everything runs the same transformations whether you use the CLI, the Kiro Power, the agent plugin or the VS Code extension.
Docs:
Announcements and blogs:
Tooling:
I am running continuous modernization against a deliberately broken polyglot repo at the moment to see what it actually catches versus what it claims. That is the next post.
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article