Praxis is the result of a decade of my experiences and frustrations working with IaC and infrastructure management in general. It aims to solve the core needs like templating, state management, continues reconciliation, and lifecycle management without needing a complex setup like Kubernetes or bolting on additional tooling like CI pipelines.
It uses a durable execution engine (https://restate.dev) to create and manage digital twins, it means every AWS call is journaled, so a crash mid-provision resumes where it stopped instead of leaving half-applied state. Each resource is a single-writer object, so there's no locking and no state file. Continuous reconciliation is just a durable timer that re-checks every resource every five minutes and either fixes drift or reports it, your choice per resource. The whole thing runs from one Docker Compose stack.
You can declare resources in CUE templates (typed, validated, with real constraints), and there are 45 AWS drivers so far across networking, compute, storage, IAM, and monitoring. Using CUE gives us a lot of goodies like policy enforcement out of the box and our templates doc (https://github.com/shirvan/praxis/blob/main/docs/TEMPLATES.m...) explains the choice of using CUE in more detail.
It's in early development and not everything is tested on a live AWS account, so use at your own risk.