steward-watch

You can automate continuous message triage by running the steward-watch loop to monitor an inbound channel, apply deterministic SOP-based handling, execute actions, and post outcomes for audit. It manages the entire watch-respond-rewatch lifecycle autonomously.

The watch-respond-rewatch cycle is an automated message triage process that continuously monitors inbound channels, executes SOP-based actions, and reliably re-engages monitoring after actions to maintain state consistency and prevent orphan tasks.

Configure the SOP orchestrator via documented frontmatter specifying swarm, channel, changelog_channel, behavior.mode, and behavior.custom_instructions. These parameters control triage behavior, chunk creation, and orchestrator coordination for the monitoring loop.

Yes, autonomous message triage enforces safe deterministic processing with explicit stop, ack, and safety guidance to avoid duplicate processing and orphan tasks while coordinating chunk creation with the orchestrator.

Recover monitoring state through the rewatch mechanism, which reliably re-engages monitoring after actions execute. This maintains state consistency across the watch-respond-rewatch lifecycle and coordinates with orchestrator-monitor for chunk management.

Yes, the orchestrator supports autonomous, queued, and custom behavior modes for inbound message triage. Custom modes allow tailored SOP-based handling while maintaining the same deterministic processing and rewatch recovery lifecycle.