I am a bit curious what it does under the hood. Is it translating the AWS CDK-generated CloudFormation to Puiumi internal representation, or a different approach? Are asset uploads handled? AWS CDK bootstrap optionally sets up S3 bucket and various IAM roles for cross-account handling, lookup etc. Are these handled in some way? This comes back to the question from @freezing-umbrella-5483 above about CDK Pipelines.

Yes - that's a generally good summary. We are using synth to create the CFN resource tree, and then mapping each thing in that tree into Pulumi resources. We do handles assets. We do currently require that you have ore-installed the CDK bootstrap. We considered options to automate installing that, but it's a bit less natural to do it automatically from within a deployment. We do expect to be able to present a useful warning if we detect it is missing with guidance to run the bootstrap process manually. We may also do something to install it automatically in the future.

That sounds very good. For the use cases where you may want to mix CDK and Pulumi or migrate from CDK to Pulumi, relying on the bootstrap being in place should not be an issue probably. It would be nice to not have a dependency to the CDK CLI to set up the bootstrap, but not a show-stopper. Is it required to have Node.js pre-installed (to run JSII/Typescript), or is that bundled somehow?