OpenAutark launches open-source local multi-agent desktop platform
OpenAutark is an open-source desktop platform designed to run locally, with a manager agent assigning work to specialized subagents. The report says it can coordinate Anthropic, OpenAI, and local open-source models, including models driven by llama.cpp, Ollama, or vLLM. It is based on the Apache 2.0 license and supports Windows, macOS, Linux, and Android. Its pluggable extensions can connect local models to external APIs, file systems, databases, and desktop automation. Developer Stanislaus Hees said the project remains at an early stage, with the number of agents that can run concurrently limited by the device’s available memory.
OpenAutark provides an open-source way to coordinate local and hosted models without requiring continuous reliance on a single public cloud provider. Its local and private-network operation may be useful where cloud API access is restricted or sensitive data is expected to remain within an organization’s environment.
在 Perplexity 等 AI 厂商相继推出“Teammates”多智能体协作功能之际,来自奥地利萨尔茨堡的 14 岁开发者斯塔尼斯劳斯·希斯(Stanislaus Hees,隶属于 Teydra Lab)推出了完全开源且在本地运行的替代方案——OpenAutark。OpenAutark 的核心词根来源于德语 „Autark“(自给自足、独立自主)。
希斯表示,此前在使用 Claude Code 和 OpenAI Codex 时,经常遇到单一平台功能受限且难以跨生态协同的问题。由于不满于被单一 AI 厂商绑定,他决定自己动手开发 OpenAutark,构建一个支持多模型混合协同的本地多智能体桌面平台。

在当今 AI 深度嵌入生产流程的背景下,多数系统高度依赖云端封闭模型和第三方 API,面临断网停摆、数据外流和算力绑架的风险。OpenAutark 的核心用途是构建一个“开放、自给自足且具备强韧性”的开源自主 AI 代理(Agent)与自动化工作流基础设施:
#OpenAutark#Teydra Lab#Teydra-Extensions#Claude Code#OpenAI Codex#Anthropic#OpenAI#llama.cpp#Ollama#vLLM