Kestra: Open Source Orchestration & Scheduling

What is Kestra?

What it is

Kestra is an open-source, event-driven orchestration platform for data, AI, and infrastructure workflows. It operates in the Java-based ecosystem and is licensed under Apache-2.0. The project provides a unified interface for both scheduled and real-time automation, supporting declarative workflow definitions in YAML while also offering visual and AI-assisted authoring.

Kestra solves the problem of fragmented workflow management across data engineering, DevOps, and AI operations by unifying scheduling, event handling, and infrastructure orchestration into a single control plane. It replaces fragmented toolchains—such as combining Airflow for batch jobs, cron for scheduling, and custom scripts for event triggers—with a consistent, version-controlled, and scalable system that enforces Infrastructure-as-Code practices.

Key capabilities

Declarative workflow definitions using YAML with real-time syntax validation and auto-completion in the built-in editor
Three synchronized authoring modes: YAML editor, No-Code visual editor, and AI Copilot that generates or modifies flows
Native support for AI agents and RAG workflows as first-class task types, with a built-in MCP server to expose flows as tools to external agents
Event-driven and scheduled triggers defined in YAML, enabling real-time automation alongside cron-based execution
Rich plugin ecosystem with hundreds of built-in tasks for databases, cloud storage, APIs, and arbitrary-language scripts
High availability and horizontal scalability via a control plane–worker architecture using gRPC communication
Full Git integration: push/pull workflows directly from the UI, support draft versions, and integrate with CI/CD pipelines

Who uses it and how

Data engineers use Kestra to build and version-control ETL/ELT pipelines with built-in error handling, retries, and conditional branching.
AI/ML teams deploy RAG workflows and agent-driven orchestration tasks directly as Kestra flows, leveraging the built-in AI Copilot and MCP server.
DevOps and infra teams automate infrastructure provisioning, monitoring alerts, and cross-service coordination using event triggers and worker isolation across regions or VPCs.

Getting started

Deploy using the official Docker image kestra/kestra:latest with kestra server standalone for local testing or kestra server cluster for production. Install via Homebrew (brew install kestra) or use the hosted option at https://cloud.kestra.io. Workflows are written in YAML and executed via the UI, CLI, or REST API.

When to use it — and when not to

Kestra replaces paid tools like Airflow (with more native event support), Prefect (with stronger declarative and AI-native features), and custom cron-based systems (with built-in resilience and version control). It requires self-hosting a control plane (PostgreSQL or H2 database, optional S3-compatible storage for artifacts) and managing worker scaling. It is not ideal for ultra-low-latency orchestration (e.g., sub-second triggers) or environments where YAML-based declarative workflows conflict with existing imperative tooling.

project readme (upstream, from github) — read inline

Open-source orchestration platform for data, AI, and infrastructure workflows

Click on the image to learn how to get started with Kestra in 3 minutes.

🌟 What is Kestra?

Kestra is an open-source, event-driven orchestration platform for data, AI, and infrastructure workflows. It unifies scheduled and event-driven automation behind a declarative, language-agnostic interface. By bringing Infrastructure as Code best practices to your data, process, and microservice pipelines, you can build reliable workflows directly from the UI in just a few lines of YAML, let the built-in AI Copilot write them for you, or generate them straight from your coding agent (Claude Code, Cursor, Windsurf, and others) with Agent Skills.

Key Features:

Everything as Code and from the UI: keep workflows as code with a Git Version Control integration, even when building them from the UI.
Three ways to build the same flow: the YAML editor, the No-Code editor, and the AI Copilot all edit the same flow and stay in sync.
AI-native orchestration: AI Agents and RAG workflows as first-class task types, plus a built-in MCP server that exposes your flows as tools to any MCP-compatible agent.
Event-Driven & Scheduled Workflows: automate both scheduled and real-time event-driven workflows via a simple trigger definition.
Declarative YAML Interface: define workflows using a simple configuration in the built-in code editor.
Rich Plugin Ecosystem: hundreds of plugins built in to extract data from any database, cloud storage, or API, and run scripts in any language.
Intuitive UI & Code Editor: build and visualize workflows directly from the UI with syntax highlighting, auto-completion and real-time syntax validation.
Scalable: designed to handle millions of workflows, with high availability and fault tolerance. Workers talk to the control plane over gRPC, so they can run in another region or inside a restricted network.
Version Control Friendly: write your workflows from the built-in code Editor and push them to your preferred Git branch directly from Kestra, enabling best practices with CI/CD pipelines and version control systems. Save a flow as a draft to iterate without disrupting the revision your triggers are running.
Structure & Resilience: tame chaos and bring resilience to your workflows with namespaces, labels, subflows, retries, timeout, error handling, inputs, outputs that generate artifacts in the UI, variables, conditional branching, advanced scheduling, event triggers, backfills, dynamic tasks, sequential and parallel tasks, checks that validate preconditions before an execution starts, quotas that cap how many executions a flow can create in a time window, and skip tasks or triggers when needed by setting the flag disabled to true.

🧑‍💻 The YAML definition gets automatically adjusted any time you make changes to a workflow from the UI or via an API call. Therefore, the orchestration logic is always managed declaratively in code, even if you modify your workflows in other ways (UI, No-Code editor, AI Copilot, CI/CD, Terraform, API calls).

✨ What's New in 2.0

Kestra 2.0 is a major release. The highlights below are the short version. See What's New in Kestra 2.0 for the full list, and the 2.0 migration guide for breaking changes and upgrade steps.

AI Copilot is now a persistent sidebar with Ask, Edit, and Plan modes, and it reads your namespace metadata so suggestions reuse the credentials you already configured.
AI Agents and RAG workflows let you orchestrate LLM calls, vector stores, and chunking inside a flow, with token usage reported per execution.
The MCP server, the MCP Tool Trigger, and Agent Skills make your flows callable from AI agents and coding assistants.

Workflow components

ForEach and ForEachItem are replaced by a single Loop task.
Trigger conditions are now written as when, alongside new date helpers such as isWeekend(), isPublicHoliday(), isDayWeekInMonth(), and isLastWorkingDay(). See date functions.
The new subflow() function returns a subflow's outputs inline in an expression, with no dedicated task.
Inputs gain {label, value} options for SELECT/MULTISELECT, JSON Schema validation, and a FORM type that groups inputs into a multi-step wizard.
Draft revisions let you stage flow changes while triggers keep running the last published revision.
Triggers can attach labels to the executions they create.

Developer experience

The No-Code editor builds flows block by block, with a guided form per task and the available upstream outputs

Frequently asked questions

Is Kestra free to use?

Kestra is open source under the Apache-2.0 licence. There is no licence fee and no seat count — you can self-host it or, where the project offers one, pay a vendor for a managed version instead.

What does Kestra do?

Declarative workflow orchestration for data, AI, and infra

What is Kestra written in?

Kestra is primarily written in Java. Its source is publicly available at https://github.com/kestra-io/kestra, and it has 28,146 GitHub stars.