gastown

Multi-agent workflow orchestration coordinates AI workers across rigs to automate task execution from assignment to final merge. It applies cross-rig coordination using transient polecats, persistent crew, and per-rig watchers for development and deployment tasks.

Hook-driven execution triggers AI workers via GUPP-based hooks, supporting sling, convoy, and refinery triggers. This mechanism automates end-to-end task execution by coordinating polecats, crew, witnesses, and refinery to complete multi-step workflows.

Automated diagnostics and escalation handle health checks and trigger human intervention when issues arise during workflow execution. This ensures resilient multi-agent orchestration by monitoring worker pipelines and managing failure recovery automatically.

Multi-agent workflow orchestration integrates with the Beans memory system for persistence, providing workflow visibility across convoys and hooks. This allows cross-rig projects to maintain state and coordinate polecats and crew effectively throughout task execution.

Triggering a specific AI worker on a remote rig is done by slinging a bead to that rig, which initiates a polecat and activates the hook-driven workflow. This starts the orchestrated multi-step task execution process automatically.

Cross-rig orchestration is needed when multi-step tasks require coordination across multiple rigs with transient polecats and persistent crew for development, integration, and deployment. It enables scalable multi-agent workflows with resilience for complex project pipelines.