Inspecting agents

Double-click an agent in the Explorer (or select it and press ⌘O) to open it as a tab. The tab’s toolbar controls the agent, and its own tabs show the agent’s state. The dot on the tab, like the one in the Explorer, shows whether the agent is active.

Controlling an agent

Control	Effect
Pause	Stops the agent after its current cycle, so it can be run a step at a time.
Resume	Lets a paused agent run again.
Step	Runs one cycle of a paused agent.
Terminate	Ends the agent.
Code	Opens the agent’s code (see below).

The right of the toolbar shows the agent’s cycle count and state. In the history mode it also shows the event being viewed, and the controls are disabled because they act on the running agent, not the past.

Beliefs

The Beliefs tab lists what the agent currently believes.

An agent's Beliefs tab, listing the artifacts it knows about, such as accessPoint("ap1",ap1) and warehouse(warehouse).

Intentions

The Intentions tab lists the agent’s intentions, one per row: the event that started the intention and, after the arrow, the subgoal it is pursuing now and how deep its stack of subgoals is. Select an intention to see its plan below: each rule in the stack with the values its variables are bound to.

The Intentions tab with an intention selected: +!start() and its subgoals +!fulfilOrders() and +!transfer("r1","ws1","sp1"), each with its variable bindings, such as {name = "r1", from = "ws1", to = "sp1"}.

In the history mode the tab lists the intentions the agent had at the selected point; their plans and bindings are only shown live.

Events

The Events tab shows the events waiting to be handled and those the agent has recently handled. Events that no rule matched are shown in red: they often explain why an agent did nothing. Click an event to see the system just after it.

The strategy agent's Events tab: no events waiting, and its handled events with their times, such as +!makeApplePie() and the message @message("inform","r1",transferred("ap1","ws1"),[]).

Rules

The Rules tab lists the rules the agent’s program added while it was running (with S.addRule(...)). Select a rule to see its source; Remove removes it from the agent.

Messages

The Messages tab lists the messages the agent has sent and received.

The Messages tab of an agent, listing the requests and replies it has exchanged.

Artifacts

With astra-cartago, the Artifacts tab lists the observable properties the agent perceives, grouped by artifact. Compare them with the artifact’s own state (below) when an agent does not seem to notice a change.

The Artifacts tab, showing the properties the agent perceives from the artifacts it is focused on.

Viewing an artifact

Double-click an artifact in the Explorer’s Environment section to open it: its type, the agent that created it, the agents focused on it, its observable properties, and (live) its operations and running operations. The cartago.* button in the section’s heading also lists CArtAgO’s own artifacts.

The tab of the artifact ap1: an artifacts.AccessPoint created by factory, with its observable property available("none") and its operations init, install, observeProperty, retrieve and take.

Viewing an agent’s code

Code on an agent’s toolbar (or Show Code on an agent’s right-click menu) opens the agent’s ASTRA class. The code is the deployed code, read from the program’s classpath, with syntax highlighting and line numbers. The classes it extends are links in the heading: click one to open its code too.

An agent's code: BasicStrategy.astra with syntax highlighting and line numbers.

Several views at once

Right-click a tab and choose Split Right or Split Down (or press ⌘\ or ⌘⇧\) to open a second view of it in a new pane. Each view is independent, so you can watch an agent’s beliefs and its intentions side by side, or two different agents. Move to Next Pane moves a tab between panes, and a pane closes with its last tab.

The main area split into two panes, each showing the same agent: its Beliefs on the left and its Intentions on the right.