IntentCenter

Source Application A — automation and workflows

Jobs

Jobs are the primary user-defined automation mechanism: Python classes packaged with the core install, shipped via Git repositories, or provided by apps. They run on demand or on a schedule, with optional approval gates.

Capabilities

• Read and write inventory through the supported ORM and service layers
• Prompt operators for inputs via declared variables (strings, integers, booleans, object pickers, file uploads, etc.)
• Emit structured logs and status records suitable for auditing
• Target optional queues (worker pools or Kubernetes job runners) for isolation or scale

Management

Each job class has a persisted record enabling enable/disable, metadata overrides, and cleanup of stale definitions when Git sources change.

Job buttons

Jobs can be exposed as buttons on object detail screens when conditions match, shortening common operational paths.

Job hooks

Hooks trigger jobs automatically when objects matching criteria are created, updated, or deleted—event-driven automation layered on change detection.

Scheduling and approvals

• Schedules integrate with the background task system.
• Approval workflows attach to models (and job scheduling) so sensitive actions require one or more approver groups in sequence before execution proceeds.
• Denials and pending states are tracked per workflow stage.

Device interaction (NAPALM)

Optional integration with the NAPALM library lets jobs or utilities retrieve live device state (facts, interfaces, etc.) for comparison against inventory. This is presented as a bridge between documented intent and operational reality, not as a full monitoring system.

Import/export and compliance (conceptual)

Feature guides reference data compliance and golden configuration style workflows; those advanced scenarios often rely on apps built on top of core jobs and inventory. Core provides the hooks; specific compliance engines are extension territory.

Kubernetes execution

For large or isolated workloads, jobs may run as Kubernetes jobs with documented wiring between the web app, queue broker, and cluster. This is an operational deployment pattern rather than a domain concept.

This site is open source. Improve this page.