open-multi-agent
Self-hosted TypeScript agent runtime with durable approvals and verifiable run records. Own it, approve it, audit it.
What is open-multi-agent?
Self-hosted TypeScript agent runtime with durable approvals and verifiable run records. Own it, approve it, audit it.
Overview of open-multi-agent
What it is
OMA (Open Multi-Agent) is a self-hosted TypeScript agent runtime for coordinating agents, adding human approval gates, and inspecting runs offline. It is open source under the MIT license, launched 2026-04-01. No telemetry, no hosted control plane. Your keys, your models — cloud, local (Ollama, vLLM, llama-server), or Chinese providers — your environment.
Key features
- Durable approvals: pause before executing a plan, dispatching a task, or calling a tool for human approval, with persistent storage so a restarted process can continue from the reviewed content.
- Verifiable journal: enable a run journal to record model inputs, tool calls, and results for later inspection;
verifyRun()checks recorded sources and content for consistency offline. - Governance floor: declare required roles and their execution order, then check whether the actual run met those requirements; applications must check
governanceConclusionseparately. - No telemetry, no hosted control plane: a library with no OMA backend or account, making no analytics, license, update, or phone-home request.
- Built-in adapters for Anthropic, OpenAI, Azure OpenAI, Bedrock, Gemini, Grok, NVIDIA, Copilot, DeepSeek, Doubao, Hunyuan, MiniMax, MiMo, Qiniu, and local models via Ollama, vLLM, llama-server; also any OpenAI-compatible endpoint and Vercel AI SDK providers.
Use cases
Scaffold a PR review agent, security analysis agent, or teaching DAG. Coordinate agents with human approval gates. Inspect runs offline. Build agent teams for research, analysis, and other tasks where you need to own, approve, and audit the agent workflow.
Getting started
Requires Node.js 20 or newer. Scaffold a new project:
npm create oma-app@latest my-oma
Or add OMA to an existing backend:
npm install @open-multi-agent/core
Set OPENAI_API_KEY to run the example with OpenAI. See the repository README for more details.
License and pricing
MIT license. Open source (free).
Based on the project's public README and repository metadata.