Asya provides CLI tools and local testing workflows that let developers iterate without deploying to a Kubernetes cluster.

The asya CLI#

The asya command-line tool provides subcommands for common development tasks:

Command Purpose
asya flow compile Compile a Python flow to AsyncActor manifests
asya flow graph Generate a visual graph (SVG/PNG) of a compiled flow
asya mcp Debug MCP tool invocations locally

Local testing with Docker Compose#

Actors can be tested locally using Docker Compose with the Socket transport. No cloud credentials, no Kubernetes cluster, no queue provisioning. The Socket transport uses Unix domain sockets for message passing, making tests fast and deterministic.

The test hierarchy supports progressive confidence:

  1. Unit tests — test the handler function directly with pytest
  2. Component tests — single actor in Docker Compose
  3. Integration tests — multiple actors wired together in Docker Compose
  4. E2E tests — full stack in a Kind cluster (only for K8s-specific features)

Flow visualization#

The flow compiler can output a visual graph of the compiled actor pipeline. This makes it easy to verify that the compiler produced the expected routing before deploying.

asya flow compile my_flow.py --graph flow.svg

Fits Any Dev Flow#

Asya is not opinionated about your tooling. It integrates with what you already use:

  • CI/CD — standard Helm charts, works with any pipeline (GitHub Actions, GitLab CI, Jenkins)
  • GitOps — Flux, Argo CD, or any tool that syncs Helm releases from Git
  • Container images — any Dockerfile, OCI buildpacks, or pre-built images; no custom base image required
  • IaC — Crossplane for infrastructure, Terraform/Pulumi for cloud resources alongside

Future roadmap#

Planned developer experience improvements:

  • VS Code extension for handler development and flow visualization
  • Online debugging with Skaffold/Tilt for live cluster iteration
  • Shadow deployments for safe production testing
  • One-click deploy/undeploy from the CLI

Further reading#