A “LOCAL VERCEL” FOR AI-BUILT APPS · OPEN SOURCE

Deploy software to
your own computer.

Think Vercel-like deployment and operations, but the application never has to move to somebody else's compute. Your PC is the server; Minit keeps it running and manageable.

terminal · development main
$ cd my-app
$ minit deploy

✓ Detected:       web app
✓ Status:         running
✓ Compute:        this PC

$ minit ls
my-app   running   healthy   8000

The app keeps running after this terminal closes.
No application code or data was uploaded.
01

Run here

Your laptop, desktop, workstation, mini PC, or team machine provides the application compute.

02

Manage here

Health, restart, monitoring, secrets, rollback, encrypted backup, and recovery wrap around the local runtime.

03

Keep control

Code, data, application secrets, encryption keys, and the app process remain local by default.

THE MODEL

AI builds→minit deploy→your PC runs it→Minit manages it

“Local Vercel” is the quick analogy. The deeper product is a management layer for software whose runtime and source of truth stay on user-owned computers.

LOCAL AUTHORITY

Cloud convenience without cloud ownership.

The local machine is authoritative. Optional Minit cloud services are designed for aggregate operational status, fleet administration, alerts, and already-encrypted backup objects — not for hosting the application. Plaintext app code, data, raw logs, secrets, and decryption keys are outside the intended cloud-admin boundary.

Read the privacy boundary →

INSTALL

Stable sharing today. One-command local deployment on development main.

PyPI 0.1.0 provides minit run. The one-command local manager is currently on main / 0.2.0.dev0 and can be dogfooded directly from GitHub before an alpha release.

# stable PyPI release
pipx install minit-runtime

# current development main
pipx install --force "git+https://github.com/junghoonwoo-stack/minit.git@main"

# inside a common web project
minit deploy
minit ls

LOCAL MANAGEMENT

One command to deploy. One view for your local apps.

Minit safely detects common static, Python, and Node web-app patterns. If detection is ambiguous, it fails closed and asks for an explicit command. A local registry lets you manage deployed apps from another directory.

# common path
minit deploy

# explicit fallback
minit deploy --port 8000 -- python app.py

# from anywhere on this computer
minit ls
minit status my-app
minit restart my-app
minit logs my-app

TEMPORARY SHARING

Sharing is separate from deployment.

minit deploy means “keep this app running on this computer.” minit run means “temporarily make this already-running local app reachable through a public URL.” Current Quick Tunnel sharing is not the target private E2E architecture.

minit run --port 8000

PRIVATE BY ARCHITECTURE

The server should not need your app.

Minit's target invariant is that compromising Minit-operated infrastructure must not be sufficient to read protected application/data/backup content or gain privileged control of the local runtime. Remote services may retain minimal operational metadata and ciphertext. Complete zero-knowledge guarantees are not claimed until sandboxing and end-to-end remote-control/connectivity work is implemented and reviewed.

Read the threat model →

CURRENT RELEASE SECURITY

Temporary public sharing is still public.

The current stable minit run path changes the security boundary of a local app. Treat its generated URL as public and do not use the stable sharing flow for sensitive workloads. Persistent local management and stronger isolation are under active development.

Read the security policy →

MINIT

Your PC is the server.
Minit handles the operations.

Explore on GitHub ↗