bex-co/bex

What it solves

bex provides an open-source, self-hostable alternative to Platform-as-a-Service (PaaS) providers like Render, Heroku, and Railway. It eliminates the need to manually operate Kubernetes while allowing developers to maintain ownership of their infrastructure and data. Crucially, it is designed to be "AI-native," providing a first-class control plane that coding agents can use to deploy and manage applications just as easily as human developers.

How it works

bex uses a Go-based operator that watches declarative Kubernetes Custom Resource Definitions (CRDs) for apps, databases, and key-value stores. When a developer or agent defines a service (e.g., via a render.yaml blueprint), the system compiles this intent into resources. The operator then handles building the source code, pushing images to a registry, rolling out workloads, and configuring routing and TLS. It leverages Cluster API to provision the underlying machines and offers multiple interfaces—including a web dashboard, REST/GraphQL APIs, a CLI, and the Model Context Protocol (MCP)—to interact with the same core logic.

Who it’s for

• Developers who want a PaaS experience (connect repo $\rightarrow$ get URL) but prefer to self-host on their own infrastructure (e.g., Hetzner or local Docker).
• AI Agents (like Claude Code or Cursor) that need a programmable way to deploy and manage cloud resources via MCP.
• Teams looking for an open-source deployment platform with built-in support for managed databases, workers, and cron jobs.

Highlights

• AI-Native Control Plane: Native support for MCP allows coding agents to perform deployments and manage state via chat.
• Render Compatibility: Supports render.yaml Blueprints and provides a compatible CLI and API.
• Multi-Runtime Support: Deploys Dockerfiles or native projects in Go, Node.js, Python, Ruby, Rust, and Elixir.
• Comprehensive Resource Management: Handles web services, private services, background workers, cron jobs, static sites, PostgreSQL, and Valkey-compatible Key-Value stores.
• Infrastructure Flexibility: Can run on local Docker-container machines or production clusters using various provider overlays.