What is Portfork?
A short introduction to Portfork, the on-demand host that keeps the apps your AI agents build running 24/7 — even when your computer is off.
- By
- Portfork
- Published
- Tagged
- product intro
If you have used an AI coding agent for more than a few weeks, you have a drawer full of half-finished apps: dashboards, scrapers, internal tools, one-off automations. They all work. They all run on your laptop. And they all die the moment you close the lid.
Portfork fixes this by giving the apps your agents build a real place to run.
What it does
Portfork is an on-demand host. Point it at an app your agent built and it keeps that app running — with a stable URL, around the clock, even when your computer is off. You run it on your own machine with Open Source Portfork, or push the same app to Portfork Cloud.
Three things make this work:
- Lazy-start, idle-reap runtime. Apps wake on the first request and scale to zero between runs. A tool you hit twice a day costs almost nothing, because it is not running the rest of the time.
- Per-app isolation. On Portfork Cloud each app runs in its own hardware-isolated microVM — a real virtualization boundary, not a shared container runtime. No noisy neighbors, strong boundaries by default.
- Agent-controllable surface. Portfork exposes a dashboard, a CLI, and an MCP server, so the same agent that built the app can deploy, inspect, and manage it.
Two editions
The Open Source edition is free forever and AGPL-3.0. It runs entirely on your machine and hosts unlimited local apps with lazy-start and idle-reap.
Portfork Cloud adds always-on hosting: each app runs 24/7 in an isolated microVM that scales to zero, priced by node tier. You bring your own model access, so there is no markup on inference — your node tier prices compute and hosting, not your LLM usage.
Try it
The Open Source edition is shipping today. Install in one command and host the apps your agents are already building. See the pricing page for the full feature surface, or jump straight to GitHub.