agent-assistant-plan-team

Multi-agent orchestration coordinates specialized agents around a shared objective using enforced governance. The Golden Triangle adversarial collaboration model assigns Lead, Executor, and Reviewer roles to prevent single-agent failure during complex teamwork.

The Golden Triangle approach structures teamwork into Lead, Executor, and Reviewer roles to enforce adversarial collaboration. It is needed when coordinating large specialized agent teams to prevent single-agent failure and ensure reliable execution.

To activate a multi-agent team plan, invoke the /plan:team command. This triggers runtime boot rules, spawns Codex-native agents from the agents directory, and resolves team-skills locally to initialize the governance workflow.

Coordinating a 26-agent team relies on context-aware collaboration and deterministic team planning. By enforcing Golden Triangle governance, the workflow aligns expert agents around a shared objective to prevent single-agent failure.

Adversarial collaboration orchestration requires strict runtime boot rules and local skill resolution from specific directories. It is limited to environments supporting Codex-native agents and depends on predefined team plans to function correctly.