Why should each agent in a workflow leave artifacts?
Why A is the answer
So that later agents and reviewers can see decisions and handoffs
Artifacts record key decisions, handoffs and outcomes, so later agents and reviewers can follow the work.
In detail
- Artifacts let later agents and reviewers see decisions and handoffs.
- Name artifacts in a consistent way.
Use case
An illustrative scenario, not a real incident.
A reviewer traces why a later agent chose a design, using the notes each agent left. The chain of decisions is visible.
See it in different ways
stateDiagram-v2 [*] --> CheckDependencies CheckDependencies --> Sequential: one step needs another CheckDependencies --> Parallel: work is independent Sequential --> Handoff: clear artifact Parallel --> Isolated: own branch or workspace Handoff --> [*] Isolated --> [*]
flowchart LR S0["Agent 1 notes"] --> S1["Agent 2 notes"] --> S2["Reviewer reads the chain"]
Key point: Artifacts make decisions and handoffs visible.
Why the other options are wrong
- B. So that the workflow takes longer and looks more thorough to people
- Taking longer is not a goal.
- C. So that the models can be trained on them later by the provider
- Training is not why they exist.
- D. So that each pull request is at least a certain minimum size
- Size is not a goal.