Turn goals into traced multi-agent task DAGs with open-multi-agent
Decompose goals into task DAGs, delegate work to agents and tools, and inspect execution traces.
npx skills add agentskillexchange/skills --skill turn-goals-into-traced-multi-agent-task-dags-with-open-multi-agent
Use this skill when an operator wants an agent system to turn a goal into a reviewable task DAG, delegate subtasks, connect MCP tools, and inspect the execution path afterward. The workflow is to define the goal, let the framework build or execute the task graph, watch progress events, and review the rendered trace before accepting results. Invoke it instead of manually chaining prompts when the important output is the decomposed plan plus auditable execution. Do not publish it as a generic agent framework card or a replacement for human project judgment. The scope boundary is task-DAG generation, delegated agent execution, MCP tool connection, and trace inspection from the upstream open-multi-agent repo.
What this skill actually does
Inputs and prerequisites: Node.js or compatible TypeScript runtime, supported model provider credentials, optional MCP servers for external tools.
Setup notes: Clone the repository, install dependencies with the package manager documented upstream, configure model credentials and optional MCP tool connections, then run the documented task-DAG workflow example.
Source and verification boundary: use https://github.com/JackChen-me/open-multi-agent as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a MCP workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.