What the Graph view shows & how to read it

What the Graph View Shows and How to Read It

Overview

The graph view is a visual representation of a Digital Worker's structure. It shows the components that make up the worker and how they relate to each other, giving users a clear picture of what the worker is configured to do and how its parts connect. As MetaPrompter applies changes, the graph updates in real time to reflect what was added, modified, or removed.

The graph is not decorative. It is a functional navigation surface. Every node in the graph corresponds to a real, configurable component of the Digital Worker, and clicking a node takes the user directly to the settings for that component.

What the Graph Represents

Each node in the graph represents one configurable part of the Digital Worker. The full set of node types that can appear in the graph includes:

Instructions: the core behavioral definition of the Digital Worker, including its purpose, persona, constraints, and output guidance.

Trigger: the condition or event that initiates the Digital Worker's activity.

Tool: a callable capability attached to the worker that allows it to retrieve information or act on an external system.

Connector: the authenticated integration that gives a tool access to an external platform.

Memory: the retained knowledge the worker draws on across interactions.

Approval: a human-in-the-loop checkpoint that requires review or sign-off before the worker proceeds.

Output: the structured result or response the worker produces at the end of a workflow.

Not every Digital Worker will include every node type. The graph reflects the actual configuration of the specific worker being viewed.

How to Read the Graph

The graph shows the worker's structure as a set of connected nodes. Reading the graph means understanding what components are present, what state each component is in, and how they connect to form the overall workflow.

The position and connection of nodes reflect the worker's operational logic, which components feed into others, where decisions are made, and where human oversight is built in. A worker with an approval node, for example, has a point in its workflow where execution pauses for human review before continuing.

Node states provide additional information about each component. A node's current state indicates whether it is live, in draft, newly added, recently changed, flagged with a warning, or pending approval. Reading node states alongside node types gives a complete view of both the worker's structure and the current status of each part of it.

For a full reference on node states and what they mean, see Node States in Graph View and What They Mean.