Creating a run

A run is the top-level unit of work in AIgency. Creating one means describing what you want done, not writing out the individual tasks yourself — AgentDock plans those from your objective.

What defines a run

• Objective — a free-text description of the outcome you want.
• Constraints — free-text limits or requirements the work should respect.
• Completion criteria — what "done" means for this run.
• Concurrency and failure policy — see Multi-agent orchestration.
• Permission mode — see Approvals.

Workspace and project

Every run belongs to a workspace — the project folder the work happens against. A run can optionally also belong to an AIgency project, a lightweight organizational container distinct from a workspace: one project can group several runs, even against different workspace folders, and can carry its own default permission mode that new runs inherit.

Run status

A run starts as draft while you're still defining it, moves to ready once it's fully specified, and then progresses through running, paused, completed, failed, or stopped as execution proceeds. See Run history for how progress is tracked once a run starts.

Next steps

Once a run is created, AgentDock plans tasks against your objective and assigns agents to each — see Agent assignment and Task routing for how that assignment works, and Execution for how tasks actually run.