// the find
argoproj-labs/hera
Hera makes Python code easy to orchestrate on Argo Workflows through native Python integrations. It lets you construct and submit your Workflows entirely in Python. ⭐️ Remember to star!
Hera is a Python SDK that lets you define and submit Argo Workflows without writing YAML by hand — you decorate Python functions and compose them with operators like `>>` into DAGs, then submit directly to a running Argo Workflows server on Kubernetes. It's for teams already running Argo Workflows who want Python ergonomics instead of hand-rolled YAML manifests, particularly in data/ML pipeline contexts.
The `@script()` decorator approach for turning arbitrary Python functions into containerized templates is genuinely nice — no separate build step, and typed I/O support means you get real function signatures instead of stringly-typed parameters. It doesn't lock you out of GitOps: `to_yaml()` lets you keep committing YAML manifests if that's your deploy pipeline, so Hera can be a authoring tool rather than a runtime dependency. Docs coverage is unusually thorough — nearly every upstream Argo Workflows YAML example has a corresponding Python translation, which is a real asset when you're porting existing workflows.
This buys you nothing without an existing Argo Workflows deployment on Kubernetes — it's not a standalone orchestrator, so the actual adoption cost is standing up Argo first, and Hera just rides on top. The CLI (YAML↔Python conversion) is explicitly flagged as experimental and subject to change, so don't build automation around it yet. Auth is DIY: you're wiring up bearer tokens or port-forwarding yourself, with no first-class credential provider for EKS/GKE/AKS clusters. There's no local dry-run story — since workflows are submitted straight to the cluster, catching a mistake in DAG structure or parameter passing means a real submission and real pod scheduling, not a fast local check.