kvcache-ai/AgentENV resource

Health v2 · maintenance only

Flags: young

Adoption not part of the score

3330 stars · 292 forks observed · 2026-08-28

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

AgentENV (AENV) is a distributed platform for running large numbers of Firecracker microVM-based agent environments across machines, with snapshot/fork support and on-demand loading of OCI-compatible images via overlaybd. It powers agentic RL training at scale, achieving fast boot/resume, high memory overcommit, and persistence to S3-compatible storage.

Use cases

• run thousands of sandboxed agent environments across a cluster
• train LLM agents with reinforcement learning at scale
• fork a running sandbox into multiple independent environments for parallel agent workflows
• snapshot and resume microVM environments in under 100 ms
• load OCI images on demand without pre-warming every host
• overcommit guest memory to increase environment density on a host
• persist environment snapshots to S3-compatible object storage

When to choose

• you need massive-scale, isolated Linux sandboxes for agentic RL or evaluation
• you want fast snapshot, fork, and resume of microVM environments
• your image/snapshot footprint exceeds local disk and you need on-demand loading with caching
• you run on Linux with KVM and want high environment density via memory ballooning

When to avoid

• you need encrypted network traffic between components
• your workloads run on non-Linux hosts or lack /dev/kvm
• you only need simple containers rather than microVM isolation
• you need a managed cloud service rather than self-hosted infrastructure

Facets

infra-config · maturity active

container-runtime container-orchestration cloud deployment microservices security infrastructure-as-code cloud-computing microservices artificial-intelligence reinforcement-learning rust self-hosted cloud firecracker microvm sandboxing agent-environments rl-training snapshotting overlaybd ublk memory-ballooning object-storage containers ai-agents devops linux docker

Member repositories

For agents

markdown · JSON · MCP: product_card(name="kvcache-ai/AgentENV")