Turn a goal into a task DAG — at runtime
In most agent frameworks you draw the task graph by hand. open-multi-agent builds it for you from a goal.
The problem.
Hand-wiring nodes and edges works until the shape of the work changes — a new sub-task, a different order, another dependency — and then you’re editing the graph, not the goal. The orchestration is built for one problem instead of adapting to each.
How open-multi-agent does it.
You pass a goal to runTeam(). A coordinator agent decomposes it into a task DAG at runtime, runs the independent nodes in parallel, and synthesizes a typed result. The plan adapts to each goal instead of being drawn once; you stay in control with hooks (plan-ready / approval gates) and a hard maxTokenBudget that aborts before overspending.
When this fits.
This fits when the orchestration topology isn’t fixed — you’d rather describe the outcome than maintain a graph. If the topology is known and you want to lay it out explicitly, an explicit-graph framework is the more natural fit.
Taking this to production?
open-multi-agent is MIT-licensed and free to run yourself. When you need it delivered, integrated, or supported on a deadline, 元定义科技 (YuanASI) offers commercial delivery and support.