Skip to content
Sign in
  • Browse all toolsFilter by category and pricing
  • Browse categoriesWriting, code, image, audio and more
  • Compare toolsPut up to three side by side
  • Read the blogGuides and comparisons

AgentENV

Open-source infrastructure for running, snapshotting and forking agent environments across machines.

AgentENV is a distributed open-source platform for running agent environments at scale. AgentENV provides Firecracker sandboxes, incremental snapshots, environment forking and an E2B-compatible API for developers and infrastructure teams supporting agentic reinforcement learning and parallel workflows. AgentENV is available under the MIT license for self-hosted deployment.

Updated 9 Nov 20266 sections

AgentENV website screenshot

What is AgentENV?

AgentENV is an open-source distributed platform for running agent environments at scale, including environments used in agentic reinforcement learning training. AgentENV runs Firecracker environments across machines and supports diverse OCI-compatible images. The repository states that AgentENV powers agentic reinforcement learning training for Kimi K3, positioning the project around infrastructure for large-scale agent workloads rather than a hosted conversational assistant.

What AgentENV can do

AgentENV combines environment execution, snapshot management and resource reclamation. AgentENV loads images on demand through overlaybd, using local disk as a bounded cache that retains hot data and evicts cold data. This approach allows aggregate image and snapshot storage to exceed local disk capacity without requiring every host to be pre-warmed.

  • AgentENV runs Firecracker environments across multiple machines.
  • AgentENV loads OCI-compatible images on demand.
  • AgentENV snapshots memory and filesystem changes incrementally.
  • AgentENV forks running environments into independent sandboxes.
  • AgentENV persists snapshots to S3-compatible storage or a shared distributed filesystem.
  • AgentENV reclaims guest memory through memory ballooning.

How AgentENV works

AgentENV provides an installation script that installs the server and aenv CLI, with the server started as a systemd service. AgentENV also documents a Docker installation path and separate CLI installation. The CLI supports Linux and macOS on x86_64 and arm64. AgentENV generates an API key on first startup, which users supply through the aenv auth command.

AgentENV users pull an image into a named template, then start a sandbox and attach an interactive shell. The CLI also supports detached execution, command execution, pause, resume and deletion. AgentENV exposes an E2B-compatible HTTP API: users point E2B_API_URL at their server to use the standard E2B Python or TypeScript SDK without code changes.

Who AgentENV is for

AgentENV serves developers and infrastructure teams that need to operate agent environments across machines, especially for agentic reinforcement learning and parallel workflows. AgentENV supports workloads involving many different environment images, resumable execution and branching sandbox state. AgentENV also gives developers using E2B SDKs a documented path to run those clients against their own server.

  • AgentENV supports agentic reinforcement learning training environments.
  • AgentENV supports parallel agent workflows through sandbox forking.
  • AgentENV supports pausing idle environments until new work arrives.
  • AgentENV supports managing sandbox lifecycles through a CLI or compatible API.

AgentENV pricing and access

AgentENV is an open-source project available from its GitHub repository under the MIT license. AgentENV documents native installation, Docker setup and build-from-source instructions, along with Docker Compose and Kubernetes cluster deployment. The supplied repository text does not describe a commercial subscription or hosted service plan for AgentENV.

AgentENV strengths and deployment limitations

AgentENV brings snapshotting, sandbox forking, on-demand image loading and E2B compatibility into a self-hosted environment system. AgentENV also shares the host page cache across storage and memory-snapshot data and uses memory ballooning to return reclaimable guest memory. These capabilities address resource use as environments run longer, diverge and become idle.

  • AgentENV requires Linux kernel 6.8+ and /dev/kvm access for standard Firecracker execution.
  • AgentENV directs servers without standard KVM support to a separate PVM deployment guide.
  • AgentENV authenticates API requests but does not encrypt traffic.
  • AgentENV requires a trusted network or HTTPS termination to avoid sending API keys over untrusted plaintext connections.
  • AgentENV's documented Docker command runs the server in privileged mode with host device access.

AgentENV FAQ

What is AgentENV?

AgentENV is an open-source distributed platform for running agent environments at scale. AgentENV uses Firecracker environments, on-demand OCI-compatible images, incremental snapshots and sandbox forking to support agentic reinforcement learning training and parallel agent workflows.

Is AgentENV free?

AgentENV is available as an open-source project under the MIT license. AgentENV provides installation scripts and deployment documentation in its GitHub repository; the supplied product text does not list a commercial subscription or hosted service price.

Does AgentENV work with E2B SDKs?

AgentENV exposes an E2B-compatible HTTP API that works with the standard E2B Python and TypeScript SDKs. AgentENV documents pointing E2B_API_URL at the deployed server so existing SDK usage can continue without code changes.

What does AgentENV require to run?

AgentENV requires Linux kernel 6.8+ and /dev/kvm access for standard Firecracker microVM execution. AgentENV links to a PVM deployment guide for servers without standard KVM support. The separately installed CLI supports Linux and macOS on x86_64 and arm64.

Can AgentENV save and fork sandbox environments?

AgentENV incrementally snapshots memory and filesystem changes and can fork a running environment into independent sandboxes for parallel agent workflows. AgentENV persists snapshots to S3-compatible object storage or a shared distributed filesystem.

Does AgentENV encrypt API traffic?

AgentENV authenticates API requests but does not encrypt traffic. AgentENV's documentation warns against sending the API key over an untrusted plaintext network and recommends running on a trusted network or terminating HTTPS at a reverse proxy or load balancer.

From the community

No reviews yet

Be the first to review this tool. What you write helps the next person choose.

More tools like AgentENV

CubeSandbox

www.github.com

Open-source sandbox service for AI agents with MicroVM isolation, E2B SDK compatibility, and snapshot management.

Free

agent-sandbox

www.github.com

Open-source Kubernetes orchestration for stateful sandboxes, AI agent runtimes, and reinforcement learning workloads.

Free

ArcBox

www.github.com

Open-source container and VM runtime for macOS with Docker compatibility, Kubernetes and isolated AI agent sandboxes.

Free

OpenShell

www.github.com

An open-source runtime that isolates AI agents and enforces policies for files, networks, processes, and credentials.

Free