Under construction — this site is a work in progress.
pichi-vm mascot — a curled-up armadillo line drawing
pichi-vm

Container workflow
for real virtual machines

Pull and run micro-VMs over OCI registries.

  1. 01 · INSTALLInstall the Pichi binary. No host reconfiguration required.
  2. 02 · PULLPull a Pichi image from any OCI registry.
  3. 03 · RUNBoot in milliseconds.
$ apt install pichi
        
$ pichi pull ghcr.io/acme/postgres-vm:16
        
$ pichi run postgres-vm:16

Preview commands, under development.

~50 ms
boot target
OCI native
pull · push · run
Real VM boundary
hardware-isolated

How Pichi compares

Pichi boots in milliseconds like a container, with full VM isolation.

ContainersFast and portable, but shares the host kernel.Traditional VMsStrong isolation, but heavy and slow to boot.PichiContainer workflow with full VM isolation.
IsolationShared host kernelFull VM boundaryFull VM boundary
Workflowpull / push / run over OCIBespoke toolingpull / push / run over OCI
DistributionIncremental, pull only what changedShip the whole disk imageIncremental, pull only what changed
PortabilityOne image, many hostsTied to hypervisor & formatOne image for laptop, cloud, bare metal
Boot timeMillisecondsSecondsMilliseconds

Layered images, validated at runtime

Pichi images are built in layers. Every block is integrity-checked at runtime.

START

From a trusted base

Start with an image whose integrity you verify.

EXTEND

Add your own layers

Layer your changes on top. Each layer is read-only and integrity-protected.

VERIFY

Single root of trust

One anchor validates the entire stack. Any modification to a layer fails the integrity check.

Built for confidential computing

Each image defines its own hardware. Host and guest agree on resources at launch.

A minimal, stable trusted base for reproducible measurements, vs a large UEFI firmware.

Read-only and content-addressed. What you tested is what runs.