minion-orchestrator

You can coordinate long-running shell jobs and LLM subagent tasks using a Postgres-backed durable queue that ensures state persistence and clear routing policies across restarts. This provides a single observable queue for both execution lanes.

The best way to track progress and pause background workflows is to use an orchestrator that supports steering, pausing, resuming, and progress tracking for both deterministic shell tasks and open-ended subagent runs.

To submit and monitor parallel subagent work in a fan-out scenario, you route tasks through a durable queue that supports DAGs and parallel fan-out, allowing you to monitor execution across multiple flows simultaneously.

Yes, you need Postgres to ensure durability for background job queues, as it provides the persistence layer required to maintain state and visibility for shell and subagent tasks across system restarts.

Yes, you can route deterministic shell commands and open-ended LLM runs through the same queue. The orchestrator enforces safe routing policies to handle both execution lanes within a single durable workflow.