Use Open Multi Agent when an operator needs a backend agent team to decompose a goal into a task DAG, assign work to specialized agents, run independent tasks in parallel, and synthesize a final result with progress and trace visibility. This is useful for repeatable engineering or analysis workflows such as code generation, contract review, meeting summarization, competitive monitoring, and other jobs where the agent should plan and coordinate multiple roles rather than answer in a single chat turn. Invoke it instead of using a normal product UI when the workflow belongs inside a Node.js/TypeScript service or CLI and needs explicit teams, provider selection, tools, optional MCP connections, structured output, shared memory, and observable execution. The scope boundary is the concrete runTeam/runTasks operator workflow for goal-to-DAG orchestration; it is not a generic multi-agent framework listing, an SDK card, or a broad MCP product entry.