What became possible in your product with Astra that was not practical before?
Maker
Astra has made it practical for me to keep building WDL at a scope that would otherwise require a much larger engineering team.
WDL has been developed extensively with OpenAI’s flagship models across several generations — GPT-5.4, GPT-5.5, GPT-5.6 Sol, and now Astra. They have assisted with implementation, debugging, architecture, code review, testing, and changes across multiple components.
WDL is a self-hosted, multi-tenant Workers platform built on workerd, with distributed routing, Durable Objects, D1, Service Bindings, Workflows, storage integrations, deployment tooling, and multi-replica coordination.
Astra is the latest step in that progression, making it easier to reason across a large codebase and carry complex, multi-step engineering work forward. The key change is not one AI feature inside WDL, but making infrastructure software of this scope practical for a very small team.
Report
Maker
📌
Hi Product Hunt 👋
I built WDL because I wanted a self-hosted way to run Workers without giving up the platform primitives that make the Workers model useful.
WDL is an open-source, multi-tenant Workers platform built on workerd. It supports stateless Workers, Durable Objects, D1, Service Bindings, Assets, and Workflows, and is designed to run across multiple replicas on infrastructure you control.
Building it turned out to be much more than embedding workerd. Stateful routing, ownership, failover, storage consistency, service bindings, and multi-replica operation all need platform-level coordination.
The goal is not to clone Cloudflare’s entire infrastructure, but to make the Workers programming model practical for self-hosted and private-cloud environments.
I’d love feedback, especially from anyone working with serverless platforms, workerd, or self-hosted developer infrastructure.