open-multi-agent/open-multi-agent
TypeScript AI agent orchestration framework with dynamic workflows. Describe the goal, not the graph: a coordinator plans the task DAG at runtime and runs it on any LLM (Claude, ChatGPT, Gemini, DeepSeek, or local models).
What it solves
Open Multi-Agent (OMA) 提供了一種編排多個 AI agent 的方式,無需開發者手動硬編碼工作流的每一步。它允許動態規劃,任務圖會在運行時根據目標生成,同時仍能保持確定性的控制、可檢查性,以及在安全、隔離或本地環境中運行的能力。
How it works
OMA 使用協調器將高層級請求分解為運行時的任務有向無環圖 (DAG)。接著,由一個代理團隊執行該計劃。該框架提供了預覽、批准和重放計劃的工具,以及確保可靠性的多代理共識機制。它是為 TypeScript 後端構建的,可以集成到 Node.js 應用程序中,支持雲端模型(如 GPT 和 Claude)與本地開源模型混合使用。
Who it’s for
對於需要動態運行時規劃、嚴格執行控制,以及能夠在自己的基礎設施(本地或離線)上運行的 AI agentic workflows 的 TypeScript 開發者和團隊。
Highlights
- Dynamic Planning: Decomposes goals into task DAGs at runtime rather than using hand-wired graphs.
- Self-Hosted: Runs in your own environment, supporting offline and air-gapped setups with default-deny tools and auto-redacted secrets.
- Model Agnostic: Mixes cloud and local models on a single team, including a fallback parser for local models that handle tool calls as text.
- Observability: Includes an offline Run Viewer for inspecting and replaying traces, with optional OpenTelemetry integration for enterprise monitoring.
- Deteriminstic Control: Features plan preview/approval, plan freezing/replay, and multi-agent consensus mechanisms.
- Dynamic Planning: 將目標分解為運行時的任務 DAG,而非使用手動編寫的圖。
- **Self-}
\n
- Model Agnostic: 混合雲端與本地模型於單一團隊中,支持連接器-fallback-parser-for-local-models-that-handle-tool-calls-as-text.
- Observability: 包含一個離線 Run Viewer 用於檢查和重放 Traces,並可選擇性地集成 OpenTelemetry 以進行企業級監控。
- Deteriminstic Control: 提供計劃預覽/批准、計劃凍結/重放,以及多代理共識機制。
<|thought|>```json [{