TencentCloud/CubeSandbox

Health v2 · maintenance only

Flags: young no_license

Adoption not part of the score

11354 stars · 1064 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

CubeSandbox is a high-performance sandbox service for AI agents built on RustVMM and KVM, creating hardware-isolated MicroVM sandboxes in under 60ms with ~5MB memory overhead. It is E2B SDK-compatible, supports snapshots/rollback, eBPF-based network isolation, and scales from single nodes to multi-node clusters.

Use cases

• run untrusted AI agent code in isolated sandboxes
• self-host an E2B-compatible code execution sandbox
• give LLM agents a secure environment to execute Python
• spin up thousands of lightweight microVMs per server
• snapshot and roll back agent execution state
• deploy a sandbox cluster for AI workloads
• replace E2B Cloud with a self-hosted alternative

When to choose

• you need hardware-level isolation for AI agent code execution
• you want E2B compatibility without cloud dependency
• you need very fast sandbox startup and high-density deployment
• you require snapshot, rollback, or forking of running sandboxes

When to avoid

• you only need simple process-level isolation without kernel separation
• you cannot run on Linux hosts with KVM access
• you need a fully managed cloud service without self-hosting
• your host lacks glibc >= 2.31 or XFS support

Facets

service · maturity active

security container-runtime microservices deployment agent-framework networking security cloud-computing infrastructure-as-code self-hosted developer-tools go rust self-hosted cloud microvm sandbox firecracker-alternative e2b-compatible kvm rustvmm ai-agent-infrastructure snapshot-rollback ebpf code-execution ai-agents linux docker kubernetes gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="TencentCloud/CubeSandbox")