What problem does it solve? When a task is fanned out to multiple agents or subagents running in parallel, shared-file races, vague scopes, and unverified results silently corrupt the work. This Skill injects a set of coordination rules that keep parallel dispatch safe and auditable. ## Core Features & Use Cases - Dispatch boundary enforcement: Ensures parallel units never write the same file, keeping shared-file edits in the orchestrator and serializing only unavoidable overlaps. - Bounded unit briefs: Requires every dispatched unit to receive an explicit scope — what it owns, what it must not touch, and what done looks like. - Result reconciliation and verification: Treats subagent summaries as claims to check against actual diffs, and requires logging of any work dropped or deferred between waves. - Use Case: While implementing a feature split across three parallel agents, apply these rules to confirm the units touch disjoint files, collect their outputs, and reconcile conflicts before declaring the wave complete. ## Quick Start Apply the orchestration skill to plan and verify a parallel dispatch of these implementation tasks across multiple subagents.
