Core Concepts

Ordine is built around a small set of composable primitives. Understanding these concepts is key to building effective automation workflows.

Concept Map

Object ──► Operation ──► Pipeline ──► Rule
               │             │
               └── Skill     └── Job (execution)

Objects

Objects are the things that pipelines operate on — the inputs that flow through the graph. Ordine supports several object types out of the box:

Object Type	Description
folder	A directory tree — all files under a path
code-file	A single source file
github-project	A GitHub repository

Objects are the entry points of a pipeline. They appear as source nodes in the DAG and their content flows downstream to operations for processing.

Operations

An Operation is the fundamental building block — an atomic task with a configured executor backend.

Operation = Executor Config + Input Schema + Output Schema

Executor types:

• agent — AI agent (Claude or Codex) with configurable system prompt and tools
• script — Custom script execution

Agent modes:

• skill — Uses a registered skill for structured AI execution
• prompt — Direct prompt-based AI execution

Pipelines

A Pipeline composes operations into a multi-step workflow, represented as a directed acyclic graph (DAG) of typed nodes connected by edges.

Node Types

Type	Description
folder	Directory input — reads a folder tree
code-file	Single file input
operation	Executes an operation
output-local-path	Writes output to a local directory
compound	Groups multiple nodes
condition	Conditional branching
github-project	GitHub repository input

Execution Model

1. Nodes are organized into execution levels (topological sort)
2. Nodes at the same level run in parallel
3. Data flows along edges from parent to child nodes
4. Each node produces a NodeCtx (content + inputPath) for downstream consumers

Skills

A Skill is a pluggable AI agent capability that operations can reference. Skills define:

• A unique ID and label
• A description of what the skill does
• A category for organization

Skills decouple "what the agent knows" from "what the operation does", making both independently reusable.

Rules

A Rule defines when and how pipelines are triggered automatically. Rules connect events (e.g., code changes, schedules, manual triggers) to pipeline execution.

Jobs

A Job tracks the execution of a pipeline run. It includes:

• Status tracking (queued → running → done/failed)
• Real-time traces and logs
• Structured output (JSON result)
• Duration and timing metadata

Plugins

Plugins extend Ordine with domain-specific capabilities. A plugin can introduce:

• New entity types (e.g., Best Practices with checklists)
• Specialized operations
• UI extensions
• Custom triggers

The built-in code quality plugin is the first example of this model.