multi-agent-orchestrator

To orchestrate a multi-skill automation pipeline with approval gates, define an explicit DAG structure with clear input/output paths and use built-in handoff artifacts like APPROVAL.md to control execution. This framework coordinates cross-domain chains deterministically.

A DAG-style workflow for cross-domain chains coordinates sequential tasks—such as research to product to marketing—using directed acyclic graphs. It maps explicit inputs and outputs to ensure deterministic handoffs and reduce coordination errors across multiple skills.

You track multi-day execution state in a workflow pipeline by using a centralized run log and persistent session state. Artifacts like RUNLOG.md and pipeline-state.md provide auditable records to support multi-step chains over extended periods.

To start a DAG pipeline run for cross-skill orchestration, copy the provided templates into a dated workspace directory such as drafts/orchestration/<YYYY-MM-DD>-<run-slug>/. Follow the steps in dag.md to execute the first skill.

Yes, you can trigger workflow pipelines via cron or webhooks. This flexible triggering capability, combined with persistent session state, supports automated and controlled execution for multi-day or multi-step automation chains.

You should not use a DAG approach for automation orchestration when your tasks require non-deterministic, highly dynamic branching rather than explicit, pre-defined input/output paths. DAGs are designed for controlled, auditable execution with clear handoff gates.