back to microsandbox

use case / regulated workloads

inspect the runtime before you trust the workload.

your auditors can't approve a black box. microsandbox is apache 2.0. read the code that enforces the boundary, run it on your own hardware, and bring your security review real answers.

dithered padlock representing an inspectable workload boundary

01

start with something you can actually read

the isolation boundary is open source. your review starts from code and measured behavior, not vendor assurances.

microsandbox runs untrusted code in hardware-isolated microVMs with their own kernel. It is open source under Apache 2.0.

02

evaluate on your terms

run it locally with networking disabled. no account, no callback, no data leaving the building while you decide.

choose the image and workload yourself, then inspect the runtime behavior on machines your organization controls.

03

the boundary comes with you

the same runtime runs on a laptop, in our cloud, or inside your own environment. contact us to evaluate running microsandbox inside your own environment.

one sdk. one api. local or cloud is a config change, not a rewrite.

04

make data paths explicit

by default, sandboxes can reach the public internet. private, host-local, link-local, and metadata destinations are blocked. egress can be reduced to an allowlist or disabled entirely. in the cloud, the non-public block cannot be lifted, even by you.

credentials are destination-bound. the sandbox works with a placeholder; the real value is substituted host-side into supported intercepted outbound requests, only for destinations on your allow-list. an approved destination still receives the real credential and could reflect or misuse it, so scope your allow-lists accordingly.

how it works

give the review real inputs

01

read the runtime

inspect the source that creates the microvm and applies its workload boundary.

02

run it on your hardware

use your own image and disable networking while you test the workload locally.

03

make your own decision

bring the code and measured behavior to the security owner who decides whether the deployment fits.

see it work

run a local image with networking disabled

bash
msb run registry.internal/analysis:2026-08 \
  --name analysis \
  --memory 2G \
  --no-net \
  -- python /work/report.py
msb rm --force analysis

know the limits

  • no disconnected-host or byoc availability claim is made on this page.
  • contact us is an evaluation path, not a self-serve deployment promise.

evaluate the alternatives

download the runtime and start your review today.

accreditation decisions stay with your organization. we give your review real inputs, not shortcuts.