Skip to content

Glossary

This glossary defines the core concepts used throughout the Engineering Runtime documentation.

Engineering Runtime

A deterministic execution platform for engineering operations.

Engineering Runtime provides a common way to execute engineering work for humans, CI/CD systems, and AI.

It is responsible for authentication, Runtime Context, policy evaluation, capability execution, and auditing.

Engineering Capability

A reusable engineering operation executed by the Engineering Runtime.

Examples include:

  • Repository Inventory
  • Kubernetes Pod List
  • Terraform Module Update
  • Cloud Resource Report
  • Security Scan Report

Engineering Capabilities represent engineering intent rather than implementation details.

Capability Package

A versioned package containing one or more Engineering Capabilities.

Capability Packages enable the Engineering Runtime to discover, load, and execute engineering capabilities.

The package defines capabilities.

The runtime executes them.

Runtime Command

A deterministic command exposed by the Engineering Runtime.

Runtime Commands invoke Engineering Capabilities through a consistent execution model regardless of whether the request originates from a human, CI/CD system, or AI.

Runtime Context

The execution context in which an engineering operation runs.

Examples include:

  • Cloud Project
  • Kubernetes Cluster
  • Namespace
  • Repository
  • GitHub Organization

Runtime Context determines where an engineering operation is executed.

Runtime Home

The local directory managed by the Engineering Runtime.

Example:

~/.engineering-runtime/

Typical contents include:

  • Configuration
  • Policy
  • Runtime Logs
  • Cache
  • Runtime Metadata

Runtime Home is owned and managed by the Engineering Runtime.

Bootstrap

The initialization process that prepares the Engineering Runtime before execution.

Bootstrap is responsible for:

  • Creating the Runtime Home
  • Loading configuration
  • Loading policies
  • Validating runtime state
  • Preparing the execution environment

Bootstrap prepares the runtime.

It does not perform engineering work.

Runtime Agent

A lightweight software component that hosts or interacts with the Engineering Runtime.

A Runtime Agent provides a consistent interface for executing Engineering Capabilities while enforcing authentication, policy, and auditing.

Policy

A collection of rules that determine whether an engineering operation is allowed.

Examples include:

  • Allowed Runtime Commands
  • Approved Binaries
  • Environment Restrictions
  • Execution Permissions

Policy is evaluated before execution begins.

Deterministic Execution

The ability to produce the same engineering outcome for the same request regardless of the executor.

Whether initiated by a human, CI/CD system, or AI, the same Runtime Command produces the same engineering outcome.

Deterministic execution removes variability from engineering operations.

Platform-Native Authentication

The practice of reusing authentication mechanisms already provided by engineering platforms.

Examples include:

  • GitHub Personal Access Tokens
  • Google Application Default Credentials (ADC)
  • OpenShift Login
  • Kubernetes kubeconfig
  • OIDC

Engineering Runtime validates existing platform authentication rather than replacing it.

Audit

A structured record describing an engineering operation.

Typical audit information includes:

  • Timestamp
  • Executor
  • Runtime Command
  • Runtime Context
  • Engineering Capability
  • Result
  • Duration

Every engineering operation should produce an audit record.

AI Assistant

An AI system that understands engineering intent and generates Engineering Runtime requests.

Examples include:

  • ChatGPT
  • Claude
  • Gemini
  • GitHub Copilot
  • Enterprise AI assistants

AI assists with reasoning.

Engineering Runtime performs deterministic execution.

Executor

Any system capable of initiating engineering work through the Engineering Runtime.

Examples include:

  • Human Engineers
  • CI/CD Pipelines
  • AI Assistants
  • Engineering Tools

Different executors share the same deterministic execution model.

Enterprise Agnostic

An architectural principle stating that enterprise-specific behavior should be expressed through configuration rather than runtime implementation.

The runtime remains reusable across organizations.

AI Independent

An architectural principle stating that the Engineering Runtime remains independent of individual AI providers.

AI systems become interchangeable interfaces rather than execution platforms.

Configuration

Settings that adapt the Engineering Runtime to a specific environment.

Examples include:

  • Authentication Providers
  • Capability Package Sources
  • Cloud Projects
  • API Endpoints
  • Runtime Preferences

Configuration changes runtime behavior without modifying runtime implementation.

Capability Discovery

The process by which the Engineering Runtime discovers available Capability Packages and their Engineering Capabilities before execution.

Intent–Execution Separation

An architectural principle separating engineering reasoning from engineering execution.

Humans and AI determine what should be done.

Engineering Runtime determines how it is executed.

Summary

Engineering Runtime terminology is intentionally implementation independent.

These concepts define a common architectural model for deterministic engineering execution across humans, CI/CD systems, and AI.