liveBook · Manning

Artifact

As explained in Build a Multi-Agent System (from Scratch)

An A2A artifact is a protocol object representing an output produced by an agent task.

An A2A artifact represents a task output. A Task can contain zero or more Artifacts, and each Artifact consists of one or more Parts. A Part holds text, raw bytes, a file URI, or JSON data. section 10.1.3

Figure 10.5 Simplified UML class diagram of the important A2A elements used in a task lifecycle (this is different from our typical UML class diagrams, which we use as build plans for our implementations)

Use the diagram to see how Artifacts relate to Tasks and Parts in the A2A task lifecycle. section 10.1.3

In an executor workflow, TaskUpdater publishes status updates and posts the final Artifact. On success, the Artifact is uploaded to the Task and becomes available to the client when the Task reaches a terminal state; on failure, an error is sent in a Message. section 10.3.1

Hailstone task

A client can send the instruction “Compute the Hailstone sequence starting at 6” in a Message. After the remote agent completes the task, its result is returned as an Artifact attached to the Task. section 10.1.3 section 10.3.1

How Build a Multi-Agent System (from Scratch) covers it

How deep it goes
Defines Artifact composition and its relationship to Tasks, then shows its role in executor success and result delivery.
Examples
Uses a Hailstone sequence request and an executor’s successful task result to illustrate Artifact flow.
What you can do after
Understand how a completed A2A task returns its result to a client through an Artifact.
How it connects
Connects Artifacts to A2A Tasks, Parts, Messages, AgentExecutor, and TaskUpdater.

Also explained in

Compare how 2 books explain artifact →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Build a Multi-Agent System (from Scratch). The section links point to the parts of the book it draws on.