What problem does it solve? When a main agent session splits large tasks across parallel sub-agents, results come back in inconsistent formats, making it hard to judge trustworthiness, retry failures, or chain follow-up work. This Skill standardizes how tasks are delegated via delegate_task and how workers report back. ## Core Features & Use Cases - Delegation Patterns: Rules for when to use workers, parallel vs. serial dispatch, and how to embed worker behavior norms into sub-agent context. - Standardized Handoff Template: A mandatory Markdown report format covering status, branch, measurements, verification level, changed files, decisions, risks, and suggested follow-ups. - Verification & Recovery: A four-tier verification scale (live-ui-verified to not-verified) plus a failure-mode retry policy (cap-hit, network-drop, tool-error) with a two-retry budget. - Use Case: You have a large refactor spanning multiple modules. Load this Skill, split the work into non-overlapping subtasks, dispatch them in parallel with delegate_task, then consume each handoff report to decide what to trust, retry, or send to code review. ## Quick Start Load the worker-handoff skill and split my current refactoring task into parallel sub-agents that each return a standardized handoff report.
