Axern 是一个开源基础设施平台,用于在隔离、有状态的沙箱中运行 AI 代理和代码,具备统一控制平面和多种运行时后端。

Stars

30

7 天增长

暂无数据

Fork 数

1

开放 Issue

0

开源协议

Apache-2.0

最近更新

2026-07-29

AI 仓库情报摘要
FR-AI / ANALYSIS

为什么值得关注

它提供了全面的、面向生产的沙箱平台,具有持久控制平面、运行时选择、真实数据平面访问以及带证据的代理执行,全部整合在同一资源模型下。

适合谁使用

  • 构建 AI 代理平台的团队
  • 需要隔离代码执行环境的开发者
  • 寻求多租户工作负载沙箱基础设施的平台工程师
  • 研究有状态代理沙箱的研究人员

典型使用场景

  • 运行带有轨迹和证据捕获的 AI 代理任务
  • 在隔离沙箱中执行不受信任的代码,并支持进程流和文件操作
  • 创建具有网络服务的临时开发或测试环境
  • 通过 Helm 或 Docker Compose 部署云无关的沙箱集群

项目优势

  • 高性能沙箱,支持 runc/runsc 运行时选择
  • 基于 PostgreSQL 的持久控制平面,可承受重启
  • 真实数据平面访问:进程流、HTTP 服务、SSH 终端、反向隧道
  • 从本地到集群的连续性,使用相同的服务边界

使用前须知

  • 处于 1.0 之前的积极开发阶段,不适合多租户生产环境
  • 需要 Docker、Kubernetes 或 Linux;macOS 仅通过 Docker Desktop 支持
  • 没有内置的多租户安全保证;需自行审查认证、TLS、网络策略

README 快速开始

Axern

Documentation · Quickstart · SDKs

Axern is open-source agentic infrastructure for running AI agents and code in isolated, stateful sandboxes. It is designed as a high-performance sandbox platform with one resource model for creating environments, executing processes, exposing services, attaching storage, opening reverse tunnels, observing lifecycle state, and retaining task evidence.

Axern is designed for teams that need more than a code-execution RPC: the control plane, node runtime, gateway, image path, SDKs, and agent harness share the same identity, lease, cleanup, and observability contracts.

Project status: Axern is pre-1.0 and under active development. It is suitable for evaluation and contribution, but operators should review the security and production boundaries before deploying multi-tenant workloads.

Why Axern

  • Sandbox as the primitive: runs, services, functions, coding workspaces, and agent tasks compose the same execution and lifecycle APIs.
  • Durable control plane: PostgreSQL-backed intent, placement, leases, retries, health, cleanup, and storage state remain authoritative across process or node restarts.
  • Runtime choice behind one model: runc and runsc workloads use the same public APIs; OCI and Nydus image paths converge at the node runtime. The resource model remains independent of a single sandbox backend.
  • Real data-plane access: process streams, files, archives, HTTP services, SSH-compatible terminals, and reverse TCP tunnels are explicit capabilities.
  • Agent execution with evidence: Axrun runs external agent bundles, verifies results, records trajectories and usage, and preserves typed artifacts.
  • Local-to-cluster continuity: Docker Compose, kind, and the cloud-neutral Helm chart exercise the same service boundaries.

Architecture

flowchart LR
    Client["CLI and SDK clients"] --> Gateway["gatewayd\npublic control and data edge"]
    Gateway --> Control["controld\ndurable intent and placement"]
    Gateway --> Tunnel["tunneld\nreverse TCP relay"]
    Gateway --> Node["axnoded\nsandbox executio

项目描述

Open-source sandboxes for AI agents, untrusted code execution, and durable services.

相关仓库与替代方案

根据分类、Topic 和编程语言匹配的相似项目。

deerwork-ai
精选
deerwork-ai GitHub avatar

deer-workflow

An open-source Dynamic Workflow runtime that combines deterministic TypeScript orchestration with replaceable Agent runtimes.

AI 与机器学习大语言模型
312
uzairansaruzi
精选
uzairansaruzi GitHub avatar

hermex

Hermex is a native SwiftUI iPhone app that lets you control a self-hosted Hermes AI agent directly from your phone, with no subscriptions, tracking, or third-party relay.

AI 与机器学习大语言模型
941
mereyabdenbekuly-ctrl
精选
mereyabdenbekuly-ctrl GitHub avatar

clodex-ide

Clodex is an open-source, local-first agentic IDE that combines persistent AI tasks, code, terminal, browser, Git, models, memory, and governed execution in one Electron workspace, currently in technical preview.

AI 与机器学习AI 智能体
859