K8EK8E
Home
  • What is K8E
  • Getting Started
  • Sandbox Runtime Setup
  • Sandbox CLI
  • Sandbox Examples
  • Sandbox with E2B SDK
  • DeepSeek Harness
  • Architecture
  • Components
  • Advanced Installation
  • K8E vs Alternatives
  • Contributing
Changelog
  • English
  • ็ฎ€ไฝ“ไธญๆ–‡
GitHub
Home
  • What is K8E
  • Getting Started
  • Sandbox Runtime Setup
  • Sandbox CLI
  • Sandbox Examples
  • Sandbox with E2B SDK
  • DeepSeek Harness
  • Architecture
  • Components
  • Advanced Installation
  • K8E vs Alternatives
  • Contributing
Changelog
  • English
  • ็ฎ€ไฝ“ไธญๆ–‡
GitHub
  • Guide

    • What is K8E?
    • Getting Started
    • Sandbox Runtime Setup
    • Sandbox CLI
    • Sandbox Examples โ€” Driven by Your AI Agent
    • Sandbox with the Official E2B SDK
    • DeepSeek Harness + k8e-sandbox โ€” Step-by-Step
    • Architecture
    • Components
    • Advanced Installation
    • K8E vs Other Sandbox Platforms
    • Contributing

What is K8E?

K8E is the Open Source Agentic AI Sandbox Matrix โ€” a Kubernetes-native platform for running secure, isolated AI agent workloads at scale, packaged as a single binary under 100MB.

As autonomous AI agents increasingly generate and execute untrusted code, robust sandboxing infrastructure is no longer optional. K8E ships everything needed to spin up a production-grade cluster in under 60 seconds, with first-class primitives for agent isolation, resource governance, and persistent execution environments โ€” purpose-built for the AI era.

๐Ÿ”’ One cluster. Many agents. Zero trust between them.

Quick Facts

  • k8e.sh โ€” Open Source Agentic AI Sandbox Matrix
  • A CNCF-conformant Kubernetes distribution in a single binary under 100MB
  • Purpose-built for secure, isolated AI agent execution at scale
  • Up and running in 60 seconds
  • Inspired by K3s

Install

curl -sfL https://k8e.sh/install.sh | sh -

That's it. Your agentic sandbox matrix is ready. ๐Ÿค–

Sandbox Capabilities

CapabilityDescription
๐Ÿ”’ Hardware IsolationPluggable runtimes: gVisor (default), Kata Containers, Firecracker microVM
๐ŸŒ Network PoliciesCilium eBPF toFQDNs egress control โ€” per-session, no proxy process needed
๐Ÿ E2B SDK CompatibleThe official e2b SDKs (Python/JS, unmodified) work natively โ€” create, exec, files, watch, pause/resume (guide)
๐Ÿšช Gateway API IngressThe Cilium Gateway API fronts every sandbox port โ€” HTTP :80/:443 + gRPC :50051 (mTLS passthrough)
โš–๏ธ Resource QuotasCPU/memory caps per agent session to prevent runaway costs
๐Ÿ’พ Persistent WorkspacesEvery session carries a PVC workspace (1Gi local-path by default); warm pods boot with their claim pre-provisioned โ€” fast start and durable files
๐Ÿง  Warm PoolPre-booted sandbox pods for sub-500ms session claim latency
๐Ÿ“ธ Content-Addressed SnapshotsSHA-256 CAS layerstore with zstd compression, chunked multi-layer manifests, incremental --base restore, server-side registry, autosquash
๐Ÿ“œ Exec TranscriptsFile-backed, windowed, offset-resumable command transcripts โ€” k8e-sandbox-cli log
๐Ÿ“Š ObservabilityPrometheus metrics, disk-only NDJSON event stream, process topology โ€” events / ps CLI
๐Ÿ”„ Sub-agent ReuseSub-agents share the parent pod + workspace; isolated reset
๐Ÿงพ CLI Catalog + MCPMachine-readable command surface for SDK generation โ€” catalog; native MCP server via k8e-sandbox-cli mcp-serve, also exposed through the K8E API Gateway
๐Ÿค agent-sandbox compatibleWorks with kubernetes-sigs/agent-sandbox
๐Ÿ”„ SKILL + CLIAI agents (Claude Code, Codex, Pi) connect via k8e-sandbox-cli CLI commands or the /k8e-sandbox skill

Project Links

  • ๐Ÿ› Bug Reports
  • ๐Ÿ’ก Feature Requests
  • ๐Ÿ” Open PRs
  • ๐Ÿ›ก๏ธ Security Advisories
  • ๐Ÿ“„ Apache License 2.0 โ€” see LICENSE
Last Updated: 10/24/26, 4:45 AM
Next
Getting Started