recipe-agent-team-run-lifecycle

You orchestrate a daemonless agent-team workflow by following a structured lifecycle from initialization to closure, covering run creation, worker dispatch, progress monitoring, and recovery without requiring a background daemon.

The end-to-end run lifecycle includes initialization, run and task creation, worker dispatch, monitoring blockers, recovering stalled tasks, inbox coordination, and final closure to ensure safe, repeatable execution.

Recover stalled tasks during agent-team orchestration by applying explicit step progression with guardrails for failure scenarios, allowing the orchestrator to monitor blockers and safely resume workflow execution.

Yes, agent-team orchestration enforces runtime prerequisites and metadata, including required binaries and dependent skills, to ensure safe and repeatable execution before the workflow starts.

Coordinate workers and manage state by using deterministic prompts for consistent inputs and clear handoffs, alongside explicit step progression to maintain safe state management and recoverability.

A daemonless approach may not suit workflows requiring continuous background processing or persistent event listeners, as it relies on explicit step progression and active orchestrator monitoring rather than a persistent daemon.