Google开源AX：面向AI Agent的声明式执行编排平台
Google's Open Agentic Orchestrator
Agent基建是当下热点，AX提供了从Google DeepMind沉淀的一手工程实践，解决了Agent长连接与状态管理的痛点，值得做Agent的同学关注其架构设计。
Declare an agentic task. AX runs it at scale.
声明一个智能体任务。AX 以大规模方式运行它。
AX sandboxes your task, wires up its workspace, fences its network, and helps you run billions of them per cluster. Either use a single task per agent, or compose as many as your agent needs.
AX 对您的任务进行沙箱隔离，配置其工作区，限制其网络访问，并帮助您每个集群运行数十亿个任务。您可以为每个智能体使用单个任务，也可以根据智能体的需求组合尽可能多的任务。
Get started View on GitHub
开始使用 在 GitHub 上查看
Pause
暂停
$ cat task.yaml
apiVersion: ax.io/v1alpha1
kind: Workspace
metadata:
name: golang
spec:
git:
- repo: https://github.com/golang/go.git
branch: "my-fix"
---
apiVersion: ax.io/v1alpha1
kind: Task
metadata:
name: test
spec:
workspaces:
- name: golang
goal: "Ensure that Go tool chain is available and is built from source"
debug: true
$ ax apply -f task.yaml
workspace.ax.io/golang created
task.ax.io/test created
$ ax watch task test
Watching task default/test...
[10:42:01] Phase: Pending Actor: test WorkerIP:
[10:42:05] Phase: Running Actor: test WorkerIP: 10.20.3.67
Task reached terminal phase "Running".
$ ax get tasks
NAME ATESPACE PHASE ACTOR WORKER-IP AGE
test default Running test 10.20.3.67 5s
$ ax ssh test -- ls /workspace
go
$ ax ssh test -- cd /workspace/go && go build ./...
$ ax ssh test -- ps -o pid,cmd
PID CMD
1 /usr/local/bin/ax-task-runner
12 go build ./...
$ ax ssh test -- touch notes.txt
$ ax suspend task test
task.ax.io/test suspended
$ ax resume task test
task.ax.io/test resumed
$ ax ssh test -- ls notes.txt
notes.txt
$ ax suspend task test
task.ax.io/test suspended
$ ax delete task test
task.ax.io/test deleted$ cat task.yaml
apiVersion: ax.io/v1alpha1
kind: Workspace
metadata:
name: golang
spec:
git:
- repo: https://github.com/golang/go.git
branch: "my-fix"
---
apiVersion: ax.io/v1alpha1
kind: Task
metadata:
name: test
spec:
workspaces:
- name: golang
goal: "Ensure that Go tool chain is available and is built from source"
debug: true
$ ax apply -f task.yaml
workspace.ax.io/golang created
task.ax.io/test created
$ ax watch task test
Watching task default/test...
[10:42:01] Phase: Pending Actor: test WorkerIP:
[10:42:05] Phase: Running Actor: test WorkerIP: 10.20.3.67
Task reached terminal phase "Running".
$ ax get tasks
NAME ATESPACE PHASE ACTOR WORKER-IP AGE
test default Running test 10.20.3.67 5s
$ ax ssh test -- ls /workspace
go
$ ax ssh test -- cd /workspace/go && go build ./...
$ ax ssh test -- ps -o pid,cmd
PID CMD
1 /usr/local/bin/ax-task-runner
12 go build ./...
$ ax ssh test -- touch notes.txt
$ ax suspend task test
task.ax.io/test suspended
$ ax resume task test
task.ax.io/test resumed
$ ax ssh test -- ls notes.txt
notes.txt
$ ax suspend task test
task.ax.io/test suspended
$ ax delete task test
task.ax.io/test deletedWhy AX
为什么选择 AX
Agents are a new kind of workload.
智能体是一种新型的工作负载。
They are neither microservices nor batch jobs. They accumulate state, need strict isolation, call out to model APIs and tool servers, and can burn money in a loop if nobody is watching. AX gives you four small primitives that handle all of that declaratively.
它们既不是微服务，也不是批处理作业。它们会累积状态，需要严格的隔离性，调用模型 API 和工具服务器，并且在无人监控时可能会在循环中不断消耗资金。AX 提供了四个小型原语，以声明式的方式处理所有这些情况。
Task
任务
Isolated execution
隔离执行
Run untrusted agent code in a sandbox with CPU and memory limits. Cheap to create, suspend, and throw away.
在具有 CPU 和内存限制的沙箱中运行不受信任的智能体代码。创建、挂起和丢弃的成本都很低。
Workspace
工作区
Easy workspace setup
轻松设置工作区
List the Git repos, MCP servers, and skills an agent needs, or just describe the goal. AX sets it all up in every sandbox before the task starts.
列出智能体所需的 Git 仓库、MCP 服务器和技能，或者仅描述目标。AX 会在任务启动前的每个沙箱中自动完成所有设置。
Gateway
网关
Network policies
网络策略
Define and quickly manage network policies. Lock traffic down to an explicit allowlist of hosts and ports, inject credentials to the incoming requests.
定义并快速管理网络策略。将流量锁定到明确的主机和端口白名单，并向传入请求注入凭据。
Model
模型
One place for config
统一的配置中心
Configure models, model parameters, and secrets in one place. Rotate a key or pin a new model version with one apply.
在一个地方配置模型、模型参数和密钥。只需一次应用操作，即可轮换密钥或固定新的模型版本。
How it works
工作原理
Scales up to billions of tasks.
可扩展至数十亿个任务。
AX runs on top of Agent Substrate, a compute runtime designed from the ground up for massive density and fast stateful actor lifecycles.
AX 构建在 Agent Substrate 之上，这是一个专为极高密度和快速有状态智能体生命周期而从头设计的计算运行时。
Billions of tasks
数十亿个任务
Every task runs as a lightweight actor, allowing you to scale to billions of concurrent agent sessions per cluster without orchestrator limits.
每个任务都作为轻量级智能体运行，使您能够在每个集群中将并发智能体会话扩展至数十亿个，且不受编排器限制。
Sub-second resumption
亚秒级恢复
Idle agents waiting on model responses, external tool calls, or human responses are checkpointed, suspended, and brought back in under a second with zero cold-start delay.
处于空闲状态的智能体，若正在等待模型响应、外部工具调用或人工回复，会被检查点保存并挂起，随后在不到一秒的时间内恢复运行，且无冷启动延迟。
Dense multiplexing
密集多路复用
Dozens of tasks share worker resources, turning idle waiting time into spare compute capacity so you only pay when agents are actively thinking and running code.
数十个任务共享工作器资源，将空闲等待时间转化为闲置计算能力，确保你仅在智能体积极思考和运行代码时才产生费用。
Generative platform
生成式平台
Generative features built into the platform.
平台内置生成式功能。
AX integrates generative AI directly into the platform. For example, if you want to set up a workspace just by explaining it in plain English, the environment is prepared automatically before your task starts.
AX 将生成式 AI 直接集成到平台中。例如，如果你希望通过自然语言描述来设置工作区，环境会在任务开始前自动准备就绪。
task.yaml
task.yaml
apiVersion: ax.io/v1alpha1
kind: Task
metadata:
name: data-analysis
spec:
workspaces:
- name: python-env
goal: "Set up a Python 3 development environment"apiVersion: ax.io/v1alpha1
kind: Task
metadata:
name: data-analysis
spec:
workspaces:
- name: python-env
goal: "Set up a Python 3 development environment"Generative workspaces
生成式工作区
Describe what a ready environment looks like in plain English. AX hands that goal to an agent on first boot to install toolchains and verify dependencies.
用自然语言描述所需环境的样貌。AX 会在首次启动时将这一目标交给智能体，以便安装工具链并验证依赖项。
Run anything and everything
运行任何内容
Interactive coding agents, long-running agent servers, Jupyter notebooks, headless browser testing, and custom tool runtimes—you name it.
交互式编码智能体、长时间运行的智能体服务器、Jupyter 笔记本、无头浏览器测试以及自定义工具运行时——应有尽有。
Perfect for research
研究者的理想之选
Spin up massive number of reproducible sandboxes to collect trajectories, run reinforcement learning loops, and evaluate agents at scale.
快速启动大量可复现的沙箱环境，用于收集轨迹、运行强化学习循环以及大规模评估智能体。
For builders & researchers
面向开发者与研究者
Built to be the most friendly runtime for developers and researchers.
专为让开发者和研究者获得最友好体验而构建的运行环境。
We want to make dealing with agentic infrastructure easier so you can focus on your work. AX is designed with an uncompromising focus on ergonomics, rapid iteration, and joyful workflows for both application developers and AI researchers.
我们希望简化智能体基础设施的使用，让你能专注于本职工作。AX 在设计上毫不妥协地注重人体工学、快速迭代以及愉悦的工作流，既服务于应用开发者，也服务于 AI 研究人员。
We aim to keep the runtime minimal and lightweight, while tastefully adding the essential features everyone needs to build, evaluate, and scale agents.
我们致力于保持运行环境的极简与轻量，同时恰当地添加构建、评估和扩展智能体所需的核心功能。
About
关于
Born from research, built for production.
源于研究，为生产而生。
AX was born at Google when agentic runtime systems research met frontier compute. Over years of building and operating agentic execution engines, teams across Google recognized that agentic workloads represent an entirely new computing paradigm: stateful, bursty, long-running actors that compute intensely for a minute and then wait for model responses, tool responses, or human approval. Traditional orchestrators built for stateless microservices or predictable batch jobs become cost-prohibitive when keeping idle sandboxes running, yet lack native support for sub-second suspend and resume.
AX 诞生于 Google，是智能体运行时系统研究与前沿算力的结合。在多年构建和运营智能体执行引擎的过程中，Google 各团队认识到，智能体负载代表了一种全新的计算范式：它们是状态化的、突发性的、长时间运行的实体，会进行高强度的计算（持续一分钟），然后等待模型响应、工具响应或人工批准。传统为无状态微服务或可预测的批处理作业设计的编排器，在维持空闲沙箱运行时成本高昂，且缺乏对亚秒级暂停和恢复的原生支持。
Drawing on agentic runtime research from Google DeepMind alongside deep experience in large-scale isolation, resumption, and scheduling, AX is being built as an open, declarative control plane purpose-built for agent execution. It abstracts tasks, workspaces, network policies, and models into core primitives so developers and researchers can run massive fleets of agents without reinventing the underlying infrastructure. This project heavily relies on Agent Substrate but provides agentic abstractions and generative runtime components.
借鉴 Google DeepMind 在代理运行时（agentic runtime）方面的研究，并结合我们在大规模隔离、恢复和调度方面的深厚经验，AX 正在被构建为一个专为代理执行设计的开放声明式控制平面。它将任务、工作区、网络策略和模型抽象为核心原语，使开发者和研究人员能够在无需重新发明底层基础设施的情况下运行大规模的代理集群。该项目严重依赖 Agent Substrate，但提供了代理抽象和生成式运行时组件。
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力