Overview

Polyaxonfiles are the declarative manifests used to define reusable components and executable operations. They describe what should run, which inputs and outputs are expected, which environment should be used, and how runtime values are resolved before execution.

Use this reference when you need the exact schema, available fields, supported types, and context variables used by Polyaxon during compilation.

Specification

The specification defines the sections that make up a Polyaxonfile, including components, operations, inputs and outputs, parameters, environments, mounts, plugins, lifecycle values, and validation rules.

Start here when you are authoring or reviewing a manifest:

• Component
• Operation
• Inputs and outputs
• Parameters
• Environment

Context

The context explains how Polyaxon resolves runtime values with Jinja expressions. It covers global run metadata, user-provided params, connection schemas, and references from upstream runs.

Use the context reference when you need to inject dynamic values such as project metadata, params, artifacts, or connection information into a Polyaxonfile.

Runtimes

The runtime references define the run schemas used by components:

• Job
• Service
• Distributed runtimes
• Cluster runtimes

Orchestration

The orchestration references define DAGs, mapping, matrix search, schedules, joins, hooks, events, and builds.

Start with the orchestration schemas when you need exact fields for workflow control:

• DAG
• Matrix
• Schedules
• Build

Pipeline Helpers

The helper references define cache, concurrency, and early stopping schemas used by orchestration runtimes.

Related Docs

• Scheduling explains how workloads are routed through queues, presets, resources, approvals, and Kubernetes entities.
• Runtimes explain the job and service workload types that Polyaxonfiles can run.
• Runtime references explain distributed jobs and cluster workload types.
• CLI explains how to validate and submit Polyaxonfiles from the command line.

---