multi-agent-team

To coordinate a multi-agent team for complex tasks, this skill orchestrates a four-member team of three executors and one QA under a Leadership Council to ensure clear ownership, cross-checks, and timely delivery across code, writing, research, and design projects.

Multi-agent orchestration handles project blockers by enforcing a phase-transition state machine with timeouts, escalating unresolved blockers to a Leadership Council, and requiring user decisions in SUPERVISED mode or resolving them automatically in FULL_AUTO mode.

Yes, you can use this multi-agent workflow for code, writing, research, and design projects. It dynamically assembles a cross-domain executor team and applies governance through a three-domain Leadership Council for Planning, Execution, and Quality.

To start a multi-agent orchestration workflow, describe your complex project to the system. The Leadership Council will clarify requirements through Q&A, assemble the executor and QA team, and prompt you to select either FULL_AUTO or SUPERVISED execution mode.

In multi-agent orchestration, FULL_AUTO mode allows the team to execute tasks and resolve blockers automatically, while SUPERVISED mode escalates blockers to the user for decision-making. Both modes enforce QA validation and phase transitions before final delivery.

Multi-agent workflow ensures quality before final delivery through a dedicated QA agent that validates outputs, enforces cross-check approvals from the Leadership Council, and audits progress via project logs and a WHITEBOARD system within a 7-phase state machine.