Use when planning or executing multi-wave parallel work in any repo. Covers wave locking, task ownership, handoff rules, merge gates, and safe subagent splitting by write scope.
Use this skill when the work is large enough to justify structured multi-agent execution.
README.mddocs/quickstart.mddocs/principles.mddocs/subagent-splitting.mddocs/adoption-guide.mdtemplates/task-template.mdtemplates/handoff-template.mdParallelize by disjoint write scope, not by broad topic.
Before spawning:
Safe:
apps/web/**packages/core/**examples/pack-a/**Unsafe:
Each task should include:
Each completed task should leave behind a durable handoff that captures:
Do not call a wave complete until:
Use subagents when:
Keep work local when:
This skill is intentionally file-first. If a team later adds a database, UI, or orchestration service, those should extend the same wave, task, ownership, handoff, and merge-gate model rather than replacing it.