Kestra
kestra is an open-source, event-driven workflow orchestration and scheduling platform. Workflows are defined as YAML files stored in Git-backed namespaces and executed via a language-agnostic Docker task model — any Python, Node.js, shell, or SQL script runs in its own container. The platform provides a web UI for monitoring, triggering, and editing workflows, 500+ plugins covering databases, cloud providers, messaging systems, and SaaS APIs, and a rich trigger system (webhooks, cron schedules, file arrival, Kafka, database polling). Written in Java, the project has approximately 26,900 GitHub stars and was last updated in May 2026.
Placed in Platforms / Assess / inner.
kestra occupies a distinct position in the workflow orchestration space relative to the two tools already on this radar. Compared to Temporal (Platforms/Assess/inner), kestra is configuration-over-code: workflows are YAML rather than SDK calls, making them immediately readable and editable without language-specific knowledge. This lowers the barrier to entry significantly for data pipelines, ETL workflows, and operational automation that does not require Temporal's fine-grained programmatic retry semantics. Compared to Windmill (Platforms/Assess/inner), kestra is workflow-first with richer trigger types and a broader plugin ecosystem; Windmill is script-first with auto-generated UIs.
The three tools serve overlapping but distinct use cases: Temporal for complex business process workflows requiring durable execution and code-first control, Windmill for converting scripts into scheduled tasks with UI forms, and kestra for declarative multi-step pipelines with event-driven triggers and broad integration coverage.
Inner position reflects 27k stars (the largest community of any orchestration tool on this radar), Docker Compose quickstart, and direct applicability to any project needing scheduled or event-triggered multi-step automation without writing orchestration SDK code. The Java runtime is the primary operational consideration for self-hosted deployments.
Trial gate: a YAML-defined workflow handling at least one real multi-step automation task (data pipeline, notification chain, or scheduled report) with at least one non-cron trigger type validated end-to-end.
