Building Workflows

This page walks through authoring a job workflow on the BizOps Engine canvas: adding steps from the action catalog, connecting edges, and configuring step data.

Before you start

You need:

1. A Project, Process, and Job (see Projects, Processes, and Jobs)
2. Access to Job details for that job
3. An action catalog available from the API (or local mock catalog for development)

Authoring flow

START

Open Job

Enter Job details for the target version and environment

Add Steps

Drag actions from the Step Palette onto the canvas

Connect Edges

Link outputs to inputs so execution order is clear

Configure

Fill action parameters and catalog tables for each step

Save

Persist the workflow document for the job

END

Step Palette

The palette is driven by the action-type catalog. Actions are grouped so you can search and place them without memorizing action IDs.

Typical authoring concerns in the palette:

Concern	What you do
Find an action	Search by title or browse groups/stages
Place a step	Drag onto the canvas
Understand intent	Use catalog title and description
Stay aligned with runtime	Prefer catalog actions that your deployment has plugin bindings for

Steps

Each canvas step carries configuration aligned with the BizOps step model. Important fields include:

Field	Purpose
stepName	Display name on the canvas
actionType	Catalog type / lane (for example Source, Logic, HTTP)
action	Concrete action name (for example httpGet)
catalogVersion	Catalog version the step was configured against
actionParameters	Key/value parameters for the action
catalogTableData	Tabular configuration rows from the catalog schema
selectedAction	Optional full selected action payload for advanced editing
stepDataId	Optional pointer to externalized step data

Position on the canvas is stored with the step so layout survives reload.

Edges

Edges define how steps relate. Keep graphs readable:

• Prefer a clear left-to-right or top-to-bottom happy path
• Avoid cycles — workflows are directed graphs
• Name or arrange branches so operators can tell primary vs alternate paths
• Revisit edges after deleting steps so the graph does not leave orphans

Configuration panel

Selecting a step opens the configuration panel. Use it to:

1. Confirm or change the catalog action version when available
2. Edit action parameters required by the selected action
3. Fill catalog tables (structured row/column config)
4. Save so the step data merges back into the workflow document

Changes in the panel should be reflected on the selected canvas node after sync.

Example walkthrough: observability triage

A common demo workflow:

Order	Role	Example action
1	Source	kafkaStreamTrigger
2	Logic	aiDebugScenario
3	Remediation	killProcessAction (dry run)

Build it like this:

1. Create a project such as Platform Operations.
2. Create a process such as Observability Triage.
3. Create a job for your target environment.
4. Add the three steps from the palette and connect 1 → 2 → 3.
5. Configure each step's action parameters (topic/signal details, AI scenario settings, remediation dry-run flags).
6. Save the workflow, run the job, then open cycle history to confirm step traces.

In default mock mode this path emits deterministic logs and outputs for narration — it does not consume real Kafka or terminate processes.

Save and reload behavior

Behavior	What to expect
Save workflow	Persists metadata, steps, edges, and viewport for the job
Reload editor	Fetches the workflow and rehydrates step data refs when present
Import legacy shape	Client may convert older nodes layouts into steps/edges

Related pages

• Action Catalog
• Execution and Monitoring
• Architecture