Lume

Secure cloud sandboxes for AI agents

E2B provides secure cloud sandboxes that let AI agents execute code, run terminal commands, and interact with filesystems in isolated environments. Each sandbox spins up in ~150ms with its own OS, giving agents a safe space to run untrusted code. Supports Python, JavaScript, and any language via custom Dockerfiles. Used by AI coding assistants, data analysis agents, and code interpreters. SDK available for Python and JavaScript with a simple API for programmatic sandbox control.

Open-source dev environment management with AI integration

Daytona is secure, elastic infrastructure for running AI-generated code in isolated sandboxes. It gives agents and developer workflows programmable environments with dedicated kernel, filesystem, network, vCPU, memory, and disk, backed by OCI/Docker compatibility, SDK/API access, and under-90ms sandbox startup. The project has 72,000+ GitHub stars and is AGPL-3.0 licensed.

Local microVM sandboxes for AI agent code execution

Microsandbox provides hardware-level isolated sandboxes for AI agents to execute code safely on local machines. Using libkrun microVMs and a 320ms bare-metal Linux/KVM homepage benchmark, it offers stronger isolation than Docker containers while staying lightweight enough for dev workstations. OCI-compatible with Python and Node.js runtimes. Apache-2.0 licensed with 6.6K+ GitHub stars.

Secure sandboxed runtime for AI agent execution

NVIDIA OpenShell provides kernel-level isolation for AI agent workloads with Landlock, seccomp, and network namespace sandboxing. Announced at GTC 2026 with 17 enterprise partners including Adobe, Atlassian, SAP, and Salesforce, it offers declarative YAML policy enforcement, L7 HTTP inspection, and GPU passthrough — purpose-built to contain the blast radius when autonomous coding agents interact with filesystems and networks.

Heterogeneous CPU-GPU inference and SFT for large MoE models

Open-source framework for running and fine-tuning large Mixture-of-Experts models with heterogeneous CPU-GPU execution, optimized kernels, limited VRAM and SGLang or LLaMA-Factory integrations.

Official Kubernetes and Helm reference stack built on the vLLM inference engine

Official vLLM reference implementation for scaling the existing inference engine on Kubernetes with Helm, request routing, KV-cache offload, autoscaling and Prometheus/Grafana observability.

Distributed inference orchestration above vLLM, SGLang and TensorRT-LLM

Open-source, datacenter-scale orchestration layer that coordinates vLLM, SGLang and TensorRT-LLM across nodes with disaggregated serving, KV-aware routing, multi-tier cache management and automatic scaling.

Open-source GPU control plane for scalable AI model serving

Open-source GPU cluster manager that configures vLLM, SGLang, TensorRT-LLM or custom engines, serves models through compatible APIs, and provisions SSH-accessible GPU instances across on-premises, Kubernetes and cloud environments.

Disaggregated KV cache storage and transfer for LLM serving

Open-source infrastructure for disaggregated LLM serving that pools KV caches across prefill and decode workers, with high-performance transfer, distributed storage and integrations for vLLM and SGLang.

Reusable KV cache infrastructure for scalable LLM inference

Open-source KV cache management layer that persists, offloads and reuses model key-value caches across requests and serving engines to reduce repeated prefill work and improve inference throughput.