







Every agent runs in its own microVM or gVisor sandbox with an identity, no credentials, and no open network. Idle agents suspend, resume in under 500ms on any ready worker, and keep their memory and filesystem state.
Where Should the Agent(s) Live? – OpenComputer
Isolation models, agent placement tradeoffs, credential design, and sandbox lifecycle patterns for agentic systems.

smol-machines/smolvm
An embeddable, portable, branchable virtual machine to safely run Agents locally.
Container Use for Locally Sandboxed, Background Agents in Zed
From the Zed Blog: Run AI agents in parallel without interference using containerized environments and Git Worktrees.

Ambient Associative Memory
Most agent memory waits to be queried. Ambient memory runs on every tool call — past lessons surface on their own, no rules list required.

agentOS — A faster, lighter, cheaper alternative to sandboxes
A portable open-source operating system for agents. Stateful runtime, universal agent interface, and secure code execution. One SDK, deploy anywhere.

A thousand ways to sandbox an agent
There are a thousand ways to sandbox an agent. Okay, I lied. There are three: simulated, containers, and microVMs. Here's when to use each.

Kenton Varda on Twitter / X
Dynamic Workers are now in Open Beta, all paid Workers users have access.Secure sandboxes that start ~100x faster than a container and use 1/10 the memory, so you can start one up on-demand to handle one AI chat message and then throw it away.Agents should interact with the…— Kenton Varda (@KentonVarda) March 24, 2026
Claude on Twitter / X
With self-hosted sandboxes, you can run agents in any environment you control: your own infrastructure, or managed providers like Cloudflare, Daytona, Modal, or Vercel. pic.twitter.com/OELUCt7QRf— Claude (@claudeai) May 19, 2026

Sprites: full Linux computers for your agents
Sprites are full Linux computers designed for agents instead of people. As persistent or disposable as you want, with full environment checkpointing and restore.

microsandbox - Every agent deserves its own machine
Run lightweight microVMs locally. Programmable networking, custom filesystems, secrets that never leak.
perplexityai/SPACE
Sandboxed Platform for Agentic Code Execution: pluggable-runtime sandboxes for running agent and user code in isolation.
perplexityai/SPACE
Sandboxed Platform for Agentic Code Execution: pluggable-runtime sandboxes for running agent and user code in isolation.
Fly · Computers for agents
Sandboxes aren't enough. Give your agent a real computer and get back to building.


shepherd-agents/shepherd
A runtime substrate that turns an agent's execution into a reversible, Git-like trace, so meta-agents can observe, fork, replay, and revert any run. Couples agent and environments in a copy-on-write fork ~5x faster than docker commit, with ~95% KV-cache reuse on replay. Framework built for meta-agents to supervise, optimize, and train other agents
AX
Declare an agentic task in YAML. AX sandboxes it, wires up its workspace, fences its network, and helps running it at scale.